HEALTHCARE TECHNOLOGIES SERIES 34

The Internet of Medical

Things
Other volumes in this series:

Volume 1 Nanobiosensors for Personalized and Onsite Biomedical Diagnosis

P. Chandra (Editor)
Volume 2 Machine Learning for Healthcare Technologies D.A. Clifton (Editor)
Volume 3 Portable Biosensors and Point-of-Care Systems S.E. Kintzios (Editor)
Volume 4 Biomedical Nanomaterials: From design to implementation T.J Webster
and H. Yazici (Editors)
Volume 6 Active and Assisted Living: Technologies and applications F. Florez-
Revuelta and A.A Chaaraoui (Editors)
Volume 7 Semiconductor Lasers and Diode-Based Light Sources for Biophotonics
P.E Andersen and P.M Petersen (Editors)
Volume 9 Human Monitoring, Smart Health and Assisted Living: Techniques and
technologies S. Longhi, A. Monteriù and A. Freddi (Editors)
Volume 13 Handbook of Speckle Filtering and Tracking in Cardiovascular Ultrasound
Imaging and Video C.P. Loizou, C.S. Pattichis and J. D’hooge (Editors)
Volume 14 Soft Robots for Healthcare Applications: Design, modelling, and control
S. Xie, M. Zhang and W. Meng
Volume 16 EEG Signal Processing: Feature extraction, selection and classification
methods W. Leong
Volume 17 Patient-Centered Digital Healthcare Technology: Novel applications for
next generation healthcare systems L Goldschmidt and R.M. Relova (Editors)
Volume 19 Neurotechnology: Methods, advances and applications V. de Albuquerque,
A. Athanasiou and S. Ribeiro (Editors)
Volume 20 Security and Privacy of Electronic Healthcare Records: Concepts,
paradigms and solutions S. Tanwar, S. Tyagi and N. Kumar (Editors)
Volume 23 Advances in Telemedicine for Health Monitoring: Technologies, design
and applications Tarik A. Rashid, Chinmay Chakraborty and Kym Fraser
Volume 24 Mobile Technologies for Delivering Healthcare in Remote, Rural or
Developing Regions P. Ray, N. Nakashima, A. Ahmed, S. Ro and Y. Soshino
(Editors)
Volume 26 Wireless Medical Sensor Networks for IoT-based eHealth Fadi Al-Turjman
(Editor)
Volume 29 Blockchain and Machine Learning for e-Healthcare Systems Balamurugan
Balusamy, Naveen Chilamkurti, Lucia Agnes Beena and Poongodi T (Editors)
Volume 33 Electromagnetic Waves and Antennas for Biomedical Applications
Lulu Wang
Volume 39 Digital Tools and Methods to Support Healthy Ageing Pradeep Kumar Ray,
Siaw-Teng Liaw and J Artur Serano (Editors)
The Internet of Medical
Things
Enabling technologies and emerging
applications

Edited by
Subhendu Kumar Pani, Priyadarsan Patra,
Gianluigi Ferrari, Radoslava Kraleva and Xinheng Wang

The Institution of Engineering and Technology

Published by The Institution of Engineering and Technology, London, United Kingdom
The Institution of Engineering and Technology is registered as a Charity in England &
Wales (no. 211014) and Scotland (no. SC038698).
† The Institution of Engineering and Technology 2022
First published 2021

This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publisher at the undermentioned address:

The Institution of Engineering and Technology

Michael Faraday House
Six Hills Way, Stevenage
Herts SG1 2AY, United Kingdom
www.theiet.org

While the authors and publisher believe that the information and guidance given in this
work are correct, all parties must rely upon their own skill and judgement when making
use of them. Neither the authors nor publisher assumes any liability to anyone for any
loss or damage caused by any error or omission in the work, whether such an error or
omission is the result of negligence or any other cause. Any and all such liability is
disclaimed.
The moral rights of the authors to be identified as authors of this work have been
asserted by them in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data

A catalogue record for this product is available from the British Library

ISBN 978-1-83953-273-3 (hardback)

ISBN 978-1-83953-274-0 (PDF)

Typeset in India by MPS Limited

Printed in the UK by CPI Group (UK) Ltd, Croydon
 Contents

About the editors xv

1 Internet of medical things (IoMT): a systematic review of

applications, trends, challenges, and future directions 1
M.A. Jabbar
1.1 Introduction 1
1.1.1 Internet of medical things 3
1.2 Internet of medical things (IoMT) applications 3
1.2.1 Role of IoMT during COVID-19 7
1.3 Security aspects in IoMT 9
1.4 Challenges and future directions 11
1.5 Conclusion 12
References 12

2 Non-invasive psycho-physiological driver monitoring through

IoT-oriented systems 19
Luca Davoli, Veronica Mattioli, Sara Gambetta, Laura Belli,
Luca Carnevali, Marco Martalò, Andrea Sgoifo, Riccardo Raheli and
Gianluigi Ferrari
2.1 Introduction 19
2.2 Heterogeneous driver monitoring 21
2.2.1 Wearable and inertial sensors 22
2.2.2 Camera sensors 22
2.3 In-vehicle IoT-oriented monitoring architecture 23
2.3.1 Experimental setup 23
2.3.2 HRV analysis 24
2.4 Experimental performance evaluation 26
2.4.1 Operating protocol for data collection 26
2.4.2 HR and HRV data 27
2.4.3 Experimental results 27
2.5 Conclusions and future works 29
Acknowledgments 29
References 30
vi The Internet of medical things

3 IoT-based biomedical healthcare approach 35

Anil Audumbar Pise
3.1 Introduction 35
3.2 IoT-based healthcare biomedical applications 37
3.3 IoT-based biomedical communication architecture 38
3.4 Wireless body area networks 39
3.5 RFID IoT-based biomedical communication protocol 41
3.6 Problem statement 43
3.7 IoT and biomedical healthcare system interconnection 43
3.8 Examples of IoT-based biomedical healthcare devices 43
3.8.1 Glucose monitoring 44
3.8.2 Bluetooth-enabled blood labs 44
3.8.3 Connected inhalers 45
3.8.4 Blood coagulation testing 45
3.8.5 Connected cancer treatment 45
3.8.6 Robotic surgery 46
3.8.7 IoT-connected contact lenses 46
3.8.8 A smartwatch app that monitors depression 46
3.8.9 Connected wearables 46
3.9 A summary of associated research 46
3.10 Conclusion 51
3.11 Future work 51
References 51

4 Impact of world pandemic “COVID-19” and an assessment

of world health management and economics 55
Hemanta Kumar Bhuyan and Subhendu Kumar Pani
4.1 Introduction 56
4.2 Impact on societies 57
4.2.1 Impact of social manner on COVID-19 58
4.2.2 Impact on healthcare facilities 59
4.2.3 Worldwide impact 60
4.2.4 Impact on low- and moderate-income countries 62
4.2.5 Impact on international healthcare facilities based on
medical services 62
4.2.6 Emergency caring 63
4.2.7 Constructive response reduction strategies 64
4.3 Impact of health on the emerging COVID-19 pandemic 66
4.3.1 Social determinants of health 66
4.4 Health in the clinical system 68
4.4.1 Clinical knowledge 68
4.4.2 Access to medical- and primary-care doctor 69
4.5 Role of health disease 70
4.5.1 Food deserts on cardiovascular disease (CVD) 70
 Contents vii

4.5.2 Role of food deserts on hypertension and chronic

kidney disease 70
4.5.3 Role of SDOH on obesity 71
4.6 Social setting 72
4.6.1 Determination 72
4.6.2 Active part of the community 73
4.6.3 Make surrounding foods 73
4.7 Academy 75
4.7.1 High school graduation 75
4.7.2 Psycholinguistics and literacy 75
4.8 Stable, predictable employment 76
4.9 The impact of the COVID-19 pandemic on firms 78
4.9.1 Time-saving method 78
4.9.2 Data gathering 79
4.9.3 Computations and statistical review 79
4.9.4 Statistical results 79
4.10 Healthcare, social and economic challenges in Bangladesh 81
4.10.1 Methodology 81
4.10.2 Analysis of basic healthcare 81
4.10.3 Precarious living 82
4.10.4 Social distancing 83
4.10.5 Groups with special needs 84
4.10.6 Discussion 84
4.11 Conclusions 85
References 87

5 Artificial intelligence in healthcare 93

Deepa Joshi and Anikait Sabharwal
5.1 Introduction 94
5.2 Healthcare data sources 95
5.3 Legal and ethical obstacles of artificial intelligence-driven
healthcare 96
5.4 Types of healthcare data 98
5.4.1 Structured data 98
5.4.2 Unstructured data 98
5.5 AI techniques developed for structured and unstructured data 100
5.5.1 Machine learning 100
5.5.2 Neural networks 101
5.5.3 Natural language processing 102
5.6 Major disease areas 103
5.7 Applications of AI in healthcare system 104
5.7.1 Solutions based on genetics 105
5.7.2 Development and discovery of drug 105
5.7.3 Support in clinical decisions 106
viii The Internet of medical things

5.7.4 Robotics and artificial intelligence-powered devices 106

5.8 Examples of AI used in healthcare 106
References 108

6 Blockchain in IoT healthcare: case study 111

Kripa Elsa Saji, Nisha Aniyan, Renisha P. Salim,
Pramod Mathew Jacob, Shyno Sara Sam, Kiran Victor and
Remya Prasannan
6.1 Overview 111
6.2 Existing models to secure IoT healthcare 112
6.3 Blockchain to secure IoT healthcare 115
6.3.1 Proposed system 115
6.3.2 System implementation 118
6.4 Conclusions 124
References 125

7 Adaptive dictionary-based fusion of multi-modal images

for health care applications 127
Aishwarya Nagasubramanian, Chandrasekaran Bennila Thangammal
and Vasantha Pragasam Gladis Pushparathi
7.1 Introduction 127
7.2 Learning a dictionary 129
7.3 Experimental setup and analysis 131
7.4 Overview of fusion scheme 131
7.5 Simulation results and discussion 133
7.5.1 Results on standard multimodal medical data sets 133
7.5.2 Objective analysis of standard medical image pairs 135
7.6 Summary 135
References 136

8 Artificial intelligence for sustainable e-Health 139

Shrikaant Kulkarni
8.1 Introduction 139
8.2 Reduction in margin of error in healthcare 140
8.3 EPRs, EHRs, and clinical systems 141
8.4 Barriers in an EHR system 141
8.5 Getting over interoperability issues 142
8.6 Barriers in EHR interoperability 143
8.7 UK-NHS model: characteristics and enhancements 143
8.8 MNC/MNE characteristics 144
8.9 UK-NHS model or the MNC organizational model 145
8.10 e-Health and AI 146
8.11 Sustainable healthcare 146
 Contents ix

8.12 Sustainable healthcare in the aftermath of COVID-19 147

8.13 Sustainability in staff and clinical practice during pandemic 147
8.14 Broadening health-care facilities at home during the COVID-19
pandemic 147
8.15 Sustainable development groups 147
8.16 Futuristic research directions 148
8.17 Role of AI in diabetes care—a case study 148
8.18 Artificial intelligence and its area of applications 149
8.18.1 Diagnosis 150
8.18.2 Interpretation 150
8.18.3 Monitoring 150
8.18.4 Control 150
8.18.5 Treatment planning 150
8.18.6 Drug design 150
8.19 Conclusion 151
References 152

9 An innovative IoT-based breast cancer monitoring system

with the aid of machine learning approach 155
Bichitrananda Patra, Santosini Bhutia, Trilok Pandey and
Lambodar Jena
9.1 Introduction 156
9.2 Related works 157
9.3 Proposed solution 159
9.3.1 Concept of iTBra 159
9.3.2 Dataset 159
9.3.3 Feature ranking with support vector machines 160
9.3.4 SMO algorithm 161
9.3.5 Reforms of the SMO algorithm 162
9.3.6 Breast cancer prediction proposed predictive framework 163
9.4 Study and discussion of experimental findings 165
9.4.1 Preprocessing for the dataset 165
9.4.2 Results of SVM-RFE experiment 167
9.4.3 Results of SMO classification 167
9.5 Conclusion 177
References 178

10 Patient-centric smart health-care systems for handling COVID-19

variants and future pandemics: technological review, research
challenges, and future directions 181
Adarsh Kumar, Saurabh Jain, Keshav Kaushik and
Rajalakshmi Krishnamurthi
10.1 Introduction 183
10.2 Internet of Things (IoT) network for patient-centric health-care
system 184
x The Internet of medical things

10.2.1 IoT and its applications for health-care sector 185

10.2.2 Industrial IoT (IIoT) for smart health-care system 187
10.2.3 Resourceful and resource constraint device-based
smart health-care developments 187
10.2.4 Research challenges and future direction of IoT in
patient-centric smart health-care system 187
10.3 Blockchain technology and IoT for patient-centric health-care
system 188
10.3.1 Need for the blockchain and IoT integration 189
10.3.2 Role of Internet of Things in health-care system 189
10.4 Cybersecurity and IoT for patient-centric health-care system 190
10.4.1 IoT and cybersecurity 190
10.4.2 Recent developments in ensuring confidentiality,
integrity, and availability (CIA) properties in IoT
networks for smart health-care system 192
10.4.3 IoT cyberspace and data handling processes for
information and network security 193
10.4.4 Research challenges and future direction of IoT in
patient centric smart health-care system 194
10.5 Parallel and distributed computing architecture using IoT
network for patient-centric health-care system 195
10.5.1 Cloud computing architectures using IoT network
for health-care system 196
10.6 Artificial intelligence and machine learning approaches in IoT
for smart health-care system 196
10.6.1 Healthcare, COVID-19, and future pandemic-related
datasets for smart health-care systems 199
10.6.2 Artificial intelligence in smart health-care system 199
10.6.3 Machine learning aspects in smart health-care system 205
10.7 Virtualization and IoT in IoT for smart health-care system 206
10.7.1 Operating system and storage virtualization for desktop-
and mobile-based smart health-care applications 206
10.8 IoT and quantum computing for smart health-care system 208
10.8.1 Introduction to IoT and quantum computing 209
10.8.2 Quantum computing and smart health-care system 209
10.8.3 Challenges of quantum computing to health-care data
and data security 210
10.9 Post-quantum cryptography solutions for futuristic security
in smart health-care system 210
10.9.1 Code-based cryptography for health-care data and
system 211
10.9.2 Lattice-based cryptography for health-care data and
system 211
10.9.3 Hash-based cryptography for health-care data and
system 211
 Contents xi

10.9.4 Multivariate cryptography for health-care data and

system 212
10.9.5 Supersingular cryptography for health-care data and
system 212
10.9.6 Post-quantum cryptography application and research
challenges for smart health-care system 212
10.10 Drone and robotics operation management using IoT network
for smart health-care system 213
10.10.1 Medical robots and recent developments 213
10.10.2 Drone and IoT network for health-care operations 213
10.10.3 Robot, drone, and IoT network integrations for
health-care applications 214
10.10.4 Recent developments and future directions 215
10.11 Conclusion and future scope 216
References 216

11 Application of intelligent techniques in health-care sector 225

Niharika Singh, Richa Choudhary, Thipendra Pal Singh and
Anshika Mahajan
11.1 Introduction 225
11.2 Evolution of AI in health-care informatics 226
11.3 Healthcare in India 227
11.4 Health-care dataset 228
11.4.1 Electronic medical records 229
11.4.2 Reluctant to adopt EMR/digitalization of medical
procedures 229
11.5 AI in healthcare 230
11.5.1 What could be achieved using AI in healthcare? 231
11.5.2 Challenges of using AI and possible solutions 232
11.5.3 Future scope 233
11.5.4 Current status of AI in healthcare 234
References 235

12 Managing clinical data using machine learning techniques 237

V. Diviya Prabha and R. Rathipriya
12.1 Introduction 237
12.2 Related work 238
12.3 Clinical data analysis 238
12.3.1 Dataset 1 239
12.3.2 Dataset 2 242
12.3.3 Dataset 3 244
12.3.4 Dataset 4 244
12.4 Conclusion 249
References 249
xii The Internet of medical things

13 Use of IoT and mobile technology in virus outbreak tracking

and monitoring 251
Marimuthu Narayanan Saravana Kumar, Ravi Bharath,
Balasubramani Yogeshwaran, Rajendiran Ranjith and
Krishnamoorthy Santhosh Kumar
13.1 Introduction 251
13.2 IoT in healthcare 252
13.3 IoT health-care applications 252
13.3.1 Glucose level sensing 252
13.3.2 Electrocardiogram monitoring 252
13.3.3 Blood pressure monitoring 253
13.3.4 Blood temperature monitoring 253
13.3.5 Oxygen saturation monitoring 253
13.3.6 Rehabilitation system 253
13.3.7 Medication management 254
13.3.8 Wheelchair management 254
13.3.9 Imminent health-care solutions 254
13.3.10 Health-care solutions using smartphones 255
13.4 Benefits 256
13.4.1 Simultaneous reporting and monitoring 256
13.4.2 Data assortment and analysis 257
13.4.3 Tracking and alerts 257
13.5 Challenges 258
13.5.1 Data security and privacy 258
13.5.2 Cost 258
13.5.3 Data overload and accuracy 258
13.6 Use of IoT in virus outbreak and monitoring 259
13.6.1 Using IoT to dissect an outbreak 259
13.6.2 Using IoT to manage patient care 259
13.7 Use of mobile apps in healthcare 260
13.7.1 Mobile report 260
13.7.2 Saving human resources 260
13.8 Healthcare IoT for virus pandemic management 261
13.9 Evolution of healthcare (pandemic)-based IoT 262
13.10 Increase availability of social networks 263
13.10.1 Improve efficiency of health services 263
13.10.2 Improve patients health condition 263
13.10.3 Enhances physician efficiency 264
13.11 Data privacy and security is a significant concern 264
13.11.1 Lack of information control 264
13.11.2 Digital divide among patients 265
13.12 Conclusion 265
Further reading 266
 Contents xiii

14 Video-based solutions for newborn monitoring 269

Veronica Mattioli, Davide Alinovi, Francesco Pisani,
Gianluigi Ferrari and Riccardo Raheli
14.1 Introduction 269
14.2 Vital signs monitoring 270
14.3 Video processing systems for neonatal disorder detection 271
14.3.1 Single sensor 273
14.3.2 Multiple sensors 273
14.4 Seizure detection 274
14.4.1 Performance in seizure detection 275
14.5 Apnea detection 277
14.5.1 Performance in apnea detection 279
14.6 Conclusion 279
References 280

15 IoT sensor networks in healthcare 283

Rinki Sharma
List of abbreviations 283
15.1 Introduction 284
15.2 Wireless sensor networks (WSN) and Internet of Things (IoT) 285
15.3 Role of Internet of Things (IoT) sensor networks in healthcare 287
15.4 Communication technologies for health-care IoT sensor networks 292
15.5 Challenges in the implementation of H-IoT and related research 294
15.6 Contemporary technologies to overcome the challenges on IoT
sensor networks for healthcare 297
15.6.1 Cloud/fog/edge computing for H-IoT sensor networks 297
15.6.2 Software-defined networking for H-IoT sensor networks 300
15.7 Conclusion 301
References 301

16 Machine learning for Healthcare 4.0: technologies, algorithms,

vulnerabilities, and proposed solutions 305
Saumya and Bharat Bhushan
16.1 Introduction 305
16.2 Healthcare 4.0 307
16.2.1 Main technologies of Healthcare 4.0 307
16.2.2 Health 4.0 objective 309
16.2.3 Health 4.0 application 310
16.3 Machine learning algorithms 313
16.3.1 Linear regression 313
16.3.2 Logistic regression 313
16.3.3 Support vector machines 313
16.3.4 Artificial neural network 314
16.3.5 Decision tree 314
16.3.6 Random forests 315
xiv The Internet of medical things

16.3.7 K means 315

16.3.8 Naı̈ve Bayes 315
16.3.9 Dimensionality reduction algorithm 315
16.3.10 Gradient boosting algorithm and AdaBoosting algorithm 315
16.4 Machine learning for Healthcare 4.0 316
16.5 Vulnerabilities for ML in Healthcare 4.0 318
16.5.1 Vulnerabilities in data collection 319
16.5.2 Vulnerabilities due to data annotation 320
16.5.3 Vulnerabilities in model training 320
16.5.4 Vulnerabilities in deployment phase 320
16.5.5 Vulnerabilities in testing phase 321
16.6 ML-based solutions for Healthcare 4.0 321
16.6.1 Privacy preservation 321
16.6.2 Differential privacy 322
16.6.3 Federated learning 322
16.7 Conclusion 322
References 323

17 Big data analytics and data mining for healthcare and smart city
applications 331
Sohit Kummar and Bharat Bhushan
17.1 Introduction 331
17.2 Theoretical background of smart cities 333
17.2.1 Smart people 334
17.2.2 Smart economy 334
17.2.3 Smart governance 335
17.2.4 Smart mobility 335
17.2.5 Smart environment 335
17.2.6 Smart living 336
17.3 Computational infrastructures for smart cities big data analytics 336
17.3.1 Cloud computing 338
17.3.2 Fog computing 338
17.3.3 Edge computing 339
17.3.4 Big data based on machine learning 339
17.4 Mining methods for big data 340
17.4.1 Classification 340
17.4.2 Clustering 341
17.4.3 Frequent pattern mining 342
17.4.4 Another mining method 342
17.5 Advances in healthcare sector 342
17.5.1 Big data for healthcare 343
17.5.2 Data mining for healthcare 345
17.6 Conclusion 345
References 346

Index 355
 About the editors

Subhendu Kumar Pani is the Principal at Krupajal Computer Academy, Odisha,

India. He has written over 200 scientific papers, book chapters, books, and patents.
His professional activities include roles as a book series editor (CRC Press, Apple
Academic Press, and Wiley-Scrivener), an associate editor, an editorial board
member, and/or a reviewer of various international journals. He is an associate of
several conference societies. He is a fellow of SSARSC and a life member of IE,
ISTE, ISCA,OBA.OMS, SMIACSIT, SMUACEE, and CSI.

Priyadarsan Patra is the dean and a professor of Computer Sciences at UPES

University, India. He served as a chief architect and a principal scientist at global
divisions of the Intel Corp., where he led world-class R&D for systems-on-chip and
microprocessors. His research interests include intelligent systems architecture.
Being an author of 75þ scholarly publications, 4 books, 13 international patents, a
senior member of IEEE and ACM, and an IE Fellow, Professor Patra founded IEEE
System Validation and Debug Technology Committee.

Gianluigi Ferrari is an associate professor of Telecommunications at the

University of Parma, Italy, where he coordinates the Internet of Things (IoT) Lab at
the Department of Engineering and Architecture. He has written over 400 scientific
papers/book chapters/books/patents and is an IEEE Senior Member. His research
interests include IoT, networking, and smart systems.

Radoslava Kraleva is an associate professor in the Department of Informatics,

South-West University “Neofit Rilski,” Bulgaria. She has published over 60 papers,
14 book chapters, and 3 books and is an active member in organizing international
conferences, seminars, and workshops. She is a member of the Bulgarian
Mathematicians Union and the Union of Bulgarian Scientists. Her interests include
human–computer interfaces, machine learning, pattern recognition, and the design
and development of interactive mobile applications for children.

Xinheng Wang is a professor and head of Department of Mechatronics and

Robotics at Xi’an Jiaotong-Liverpool University, Suzhou, China, with research
interests in Internet of Things (IoT), Industrial IoT (IIoT), mobile healthcare, and
acoustic localization, communications and sensing for healthcare. He was the
inventor of the first IoT product, smart trolley, in the world to deliver intelligent
services to airport passengers. He is the author/co-author of more than 200 inter-
national journal and conference papers and serves as a general chair/technical
program chair for CollaborateCom since 2018. He is the holder of 15 granted
patents. He is also an IET fellow and a senior member of IEEE.
 Chapter 1
Internet of medical things (IoMT): a systematic
review of applications, trends, challenges, and
future directions
M.A. Jabbar1

The Internet of medical things (IoMT) is rapidly changing the healthcare sector. IoMT
paired with enabling technologies like artificial intelligence, cloud computing, wireless
sensor networks can effectively monitor people’s health continuously. In this chapter,
the author provides a systematic review of the architectures of IoMT, applications, and
trends in IoMT. IoMT applications to counter the COVID-19 pandemic are also dis-
cussed. IoMT and enabling technologies will improve the quality of human lives. This
chapter addresses a few challenges while adopting IoMT in healthcare.

1.1 Introduction
In Internet of Things (IoT), all objects are considered smart objects that are paired
with each other. Sensors and actuators blend with the environment and commu-
nication is shared across the platform. As per Atzori [1] IoT can be realized in three
paradigms: (1) semantic oriented, (2) things oriented and (3) Internet oriented.
As per [2], IoT has been recognized as one of the emerging technologies that
transform the world. IoT has a complex architecture, where various components
interact with each other to find the solutions. IoT comprises cyber-physical systems
to facilitate data-driven decision process.
By integrating various sensors and data analytics, IoT offers various solutions
that can be used in smart grids, smart towns, smart homes, etc. There are three
components in IoT that enables seamless applications [3]. The three components
are (1) hardware that consists of sensors and actuators with embedded hardware, (2)
middleware used for computing and storage and (3) visualization tools.
IoT is enabled with various technologies like (1) wireless sensors networks
(WSN), (2) radio-frequency identification (RFID), (3) communication stack for
WSN, (4) hardware for WSN, (5) middleware and (6) methods for secure data
integration [3].

1
Vardhaman College of Engineering, Hyderabad, India
2 The Internet of medical things

RFID is a breakthrough technology in communication that enables in the

design of various microchips in wireless communications. Two types of RFID,
namely, active RFID and passive RFID are used in communication.
Due to the technological advancements, wireless communication have made
available for effective low-cost and low-power devices.
Hardware for WSN consists of sensors, processing units, interfaces, receivers
and power supply.
Communication stack is used to communicate with external world using the
Internet. Middleware is used to integrate service-oriented architecture and sensor
networks. Secure data aggregation methods are required to extend the lifetime of
networks.
Efficient data storage and analytics tools are needed to process the data
visualization tools that are used to help the users with the environment.
Addressing schemes are required to identify the various things that are con-
nected. These addressing schemes will help one to uniquely identify the devices
and control them with the Internet. IoT enables communication among various
machines that form a large scale wireless network [4]. IoT architecture is shown in
Figure 1.1 [5–7].
Workings of various components in the IoT are described as follows:
Sensors are used to transmit and receive data.
IoT edge devices are used to conduct processing.
Device provision helps one to connect various devices.
IoT gateway provides management, command, and control of the devices.
Stream processing is used to analyze complex execution.
Machine learning is used for predictive maintenance.
Reporting tools are used to store the data.
User management helps one to perform an action on the devices that are
connected.
Communications of IoT are illustrated in Figure 1.2. Communications of IoT
will happen in three domains, namely, (1) application domain, (2) network domain
and (3) IoT domain [8].

Cloud server
IoT edge devices

IoT sensors
Data transformation
Bulk device IoT gateway User
Stream processing
provisioning framework interface
Central management User
etc., system

Reporting Machine User

tools learning management

Things Insights Action

Figure 1.1 Architecture of IoT [5–7]

 Internet of medical things (IoMT) 3

IoT
Application Network devise
domain domain domain
Satellite

Dish Gateway

Data

Services
Person

Person
User

Figure 1.2 IoT communication [8]

IoT has been used in many applications, including agriculture, industry

environment healthcare, smart cities, and for commercial use. Applications of IoT
in various fields are shown in Table 1.1.
In addition to other segments, IoT has been used in healthcare industry. IoT
plays a vital role in healthcare industry. Applications of IoT in health sector are
shown in Table 1.2.

1.1.1 Internet of medical things

IoMT is playing an important role in healthcare industry. IoMT is an inter-
connection of various medical devices and their applications connected with
Internet.
Various elements in IoMT are listed in Table 1.3.
Patients health status recorded by IoMT will be transferred to a doctor via
cloud data centers [9,10]. IoMT supports to monitor the clinical data of patients
from hospital or home. IoMT consists of sensors and electronic circuits that take
the data from the patients and process them. General architecture of IoMT is shown
in Figure 1.3 [9,10] and the architecture of IoMT is shown in Figure 1.4 [11].

1.2 Internet of medical things (IoMT) applications

Smart healthcare one of the important aspects of human life. The use of emerging
technologies in healthcare will help one to reach the health-related services to all
the stakeholders. IoMT along with other emerging technologies is playing a vital
role in smart healthcare. Various applications of IoMT in healthcare are (1)
4 The Internet of medical things

Table 1.1 Applications of IoT

Sl. no. Reference Application

1 [33] Nursing system
2 [34] Rehabilitation
3 [35] Kidney abnormality detection
4 [36] Posture recognition
5 [37] Patient psychological condition
6 [38] Medical health monitoring
7 [39] Autistic patient monitoring
8 [40] Smart nursing
9 [41] ECG monitoring
10 [42] Smart healthcare
11 [43] Medical health band
12 [44] Smart hospital
13 [45] Monitoring sleep apnea
14 [46] M-Health
15 [47] Cardiac arrhythmia management
16 [48] Medical health management
17 [49] Medical IoT
18 [50] Health monitoring for chronic disease
19 [51] Monitoring the poultry form
20 [52] Ecological monitoring
21 [53] Water monitoring
22 [54] Home automation
23 [55] Disaster management
24 [56] Ozone mitigation challenges identification
25 [57] Greenhouse gases monitoring
26 [58] Mobile crowd sensing
27 [59] Digital forensic
28 [60] Location tracking
29 [61] Data processing
30 [62] smart home
31 [63] Smart home
32 [64] Smart home
33 [65] Vehicular monitoring
34 [66] Smart home
35 [67] Weather monitoring
36 [68,69] Smart street parking
37 [70,71] Smart city

management of chronic diseases, (2) telehealth, (3) lifestyle assessment, (4) remote
intervention, (5) drug management, (6) medical nursing system, (7) rehabilitation
system and (8) medical Bot.
This subsection gives a brief review of research related to applications of IoMT
in healthcare.
Muzammal et al. [12] presented multisensor data fusion framework. IoMT is
used for cuff-less BP measurement. In [13] authors presented a model based on
backward shortcut connections. An IoMT model is designed for emotion recognition.
 Internet of medical things (IoMT) 5

Table 1.2 Applications of IoT in health sector

Sl. no. Reference Specific application

1 [72] Disease management
2 [73] Healthcare monitoring
4 [74] Healthcare monitoring
5 [75] Medical home monitoring
6 [76] Security management
7 [77] e-Health
8 [78] Wearable device
9 [79] Boy sensors
10 [80] e-health
11 [81] Wearable device
12 [82] To counter PUEA attacks
13 [83] Biotelemetry
14 [84] Energy-efficient protocol
15 [85] Healthcare
16 [86] Healthcare
17 [87] Healthcare environment

Table 1.3 Elements that are connected in IoMT

Sl. no. Elements connected

1 Visual sensors
2 Accelerometer sensor
3 CO2 sensor
4 Temperature sensor
5 ECG sensor
6 Sensors to measure pressure
7 Gyroscope sensors
8 Humidity sensor
9 Blood pressure monitor sensor

Data fusion-enabled approach was proposed by [14]. A body sensor network is

used for heart disease prediction. A fog computing environment is used in the
proposed model. Predicting the sleep apnea using sleep apnea method was pro-
posed in [15]. Long short-term memory and convolutional neural network (CNN)
were used in the design of the proposed method.
Lin et al. [14] proposed a model for the detection of mental stress. EEG
electrodes and functional near-infrared spectroscopy were used in the proposed
method. In [16] Cabria and Gondra presented an algorithm to segment brain tumors
in MRI scans. Potential field segmentation algorithm was used based on
potential field.
Heart rate tracking using wearable sensors was suggested by [17]. Motion objects
caused by ECG and PPG data were used for to measure the performance of the
proposed model. A model to record BP from ECG sensor data was proposed in [18].
6 The Internet of medical things

PHYSICIAN AT HOME/CLINIC Hospital

Vital parameters
heart rate
heart rate variability Admin module Finance module
pulse rate
respiration rate Login management Bill to hospital Bill to patient
systolic blood pressure Physician Asset management Tarif mgmt Billing
diastolic blood pressure Users role Billing Receipt
Patient home 1 oxygen saturation
body temperature Asset allocation Patient visit
Receipt
management
Network types body mass index Application
Patient home 2 Ethernet servers
Zigbee Physician module Pharmacy module
Patient home 3 Bluetooth RESTful Patient demographic data Stock details
Wi-Fi Case history Case history
DSRC/WAVE Internet MQTT
Patient home N Diagnostics Billing
DASH 7
Procedure Receipt
6LoWPAN CoAP
Ambulance 1 Medication
LoRaWAN
Custom Network
GSM/GPRS
3G/4G server
Ambulance 2 CENTRAL NURSING STATION
Vital parameters
level of height
Hospital appliances Local
Ambulance N storage
consciousness blood Glucose level (printers, lights, air
muscular activation pain conditioners, Decision
total lung volume urine output refrigerators and so on) support
Remote healthcare Real-time monitoring

Figure 1.3 Architecture of IoMT [9,10]

LTE
IoT
Gateway

Processing Remote Medical experts

units (servers) monitoring

Communication
Sensors technologies

Figure 1.4 General architecture of IoMT [11]

Data generated from various sensors was fused and the output is fed to seven different
classifiers.
In [19] authors presented a method for emotion recognition. Fused and non-
fused physiological signal data sets were used for the evaluation of the proposed
method. A feed-forward neural network classifier is used to train the data set.
Authors in [20] modeled a recognition system for human action. The proposed
method is tested on UTD-MHAD data set.
A Gaussian filter was used to decompose the images. Baloch et al. [21] pre-
sented a method for IoT healthcare applications. The proposed method consists of
three phases to resolve issues and to maintain the efficiency. Applications of IoMT
in healthcare have been shown in Table 1.4. The illustration of IoMT applications is
shown in Figure 1.5 [22].
 Internet of medical things (IoMT) 7

Table 1.4 Research that has undertaken by adopting IoMT in healthcare

Reference Objective of the Classifier used IoMT

method
[88] Seizure detection Kriging method EEG
[89] Seizure detection Swift network classification Wearable EEG
[90] Voice recorder CNN Smart sensors
[91] Human activities CNN ECG
[92] Stroke detection K-Nearest neighbor, MLP, SVM CT

Internet of
medical things

Figure 1.5 Applications of IoMT [22]

1.2.1 Role of IoMT during COVID-19

IoMT has been deployed to stop the spread of COVID-19 pandemic. Adoption of
IoMT with other technologies like AI, big data, and block chain offers more effective
solutions. A number of researchers are developing secure IoMT applications.
IoMT supports monitoring of COVID-19 patients from their home. Various
measurements related to patients like blood pressure, heart beat will be recorded
and transferred through the cloud to the healthcare workers. This will stop the
spread of disease [23]. IoMT has been used to identify and track the origin of the
COVID-19 outbreak [24]. Taiwan’s company developed a wearable IoMT to detect
the abnormal temperatures in and send the alert notification to higher authorities
[25]. The cloud-based temperature monitoring system reduced the errors, saved
medical personal time and reduced the risk of getting infected to healthcare
officials.
8 The Internet of medical things

AI-enabled IoMT has been deployed at a Taiwan’s hospital to scan the people
coming to hospital, detect the temperature and identify whether the people wearing
the face mask or not [26]. Table 1.5 shows the various technologies used for
COVID-19. Table 1.5 shows applications of IoMT during COVID-19.
Research that is mainly focused to address the challenges faced by COVID-19 are
1. measures to enforce social distancing using the technology,
2. remote patient monitoring of the patients using IoMT,
3. IoMT-based methods for infection detection,
4. IoMT-enabled thermal screening to avoid the spread of disease.
Advance technologies to tackle the pandemic are shown in Figure 1.6 [27].
Cognitive IoT is an emerging technology that is used for an efficient utilization
of a scarce spectrum [28]. This technology is well suited to track the persons affected
with COVID-19. Cognitive IoMT (CIoMT) is a special case of IoT. IoMT has been
used in major areas to handle COVID-19 pandemic. CIoMT has been used for
1. real-time tracking of patients,
2. monitoring COVID-19-affected patients remotely,

Table 1.5 IoMT during COVID-19

Reference Application Objective of the application

[93] Tracking application Prediction of virus trend
[94] Temperature alert Risk of mortality
[95] Smart detector system Drug detector
[95] Diagnosis application Diagnosis of disease
[96] Tracing application Patient tracking
[97] Telemedicine Prediction

IoT in 5G

Cloud
computing and Advance AI and deep
big data Technologies learning
analytics

Blockchain
and Industry
4.0

Figure 1.6 Advanced technologies that are used to tackle COVID-19 [27]
 Internet of medical things (IoMT) 9

Real-time
tracking
Remote
Prevention
monitoring
and
of the
control
patient

CIoMT
Reducing
workload Rapid
of medical diagnosis
industry

Contact
Screening
tracking
and
and
surveillance
cluttering

Figure 1.7 Applications of CIoMT toward COVID-19

3. diagnosis of pandemic rapidly,

4. surveillance of COVID-19,
5. screening of patients,
6. prevention and control of pandemic disease.
Major application areas where we can use (CIoMT) are illustrated in Figure 1.7
[29].
IoMT is also effective in providing the services to orthopedic patients affected
with COVID-19. Services offered by IoMT for COVID-19 patients are
1. providing smart healthcare,
2. in terms of wearable devices,
3. patient tracking,
4. remote patient monitoring [30].
Services offered by IoMT toward COVID-19 orthopedic patients are shown in
Figure 1.8 [30].

1.3 Security aspects in IoMT

Heterogeneous physical devises that are connected to IoT constitute the attack
surface. There are many security challenges pertaining to IoT protocols and
standards.
10 The Internet of medical things

Smart
healthcare

Wearable Patient
devices tracking
IoMT facilities
for
orthopedic
patients in
COVID-19
Internet
Exigency
-based
warning
services

Remote
health
monitoring

Figure 1.8 Services offered by IoMT during COVID-19 as per [30]

Security challenges in IoT include the following:

1. identifying third party intruder,

2. diverse network update policy,
3. add in security policy,
4. physical and exposure threat [31].

Attacks on IoT classified into two types: (1) data based and (2) protocol based.
Data-based attacks include threats related to original data packets, whereas
protocol-based attacks exploit protocol-based structure. Protocol-based attacks are
again classified into two types: (1) communication protocol based and (2) network
protocol based.
Attacks will also be classified as active and passive. Category of various
attacks is shown in Figure 1.9.
IoT attacks can also be classified as shown in Figure 1.10.
Rapid evolution of IoMT also raised security problems. IoMT security
requirements and methods include (1) block chain, (2) access control methods,
(3) IDS methods, (4) authentication schemes and (5) key management methods.
Cyberattacks that happen to IoMT are listed as follows: (1) injection attacks,
(2) denial of service attack, (3) device safety and (4) data leakage attacks.
Attack surface for denial service is mostly on databases and cloud services.
Hardware and middleware are the target for device safety attacks. Injection
attacks basically target the databases. Data leakage attacks will occur to
information and network.
 Internet of medical things (IoMT) 11

Attack type

Active attack Passive attack

Masquerade Port
DOS attack Message replay Traffic sniffing
attack scanning

Figure 1.9 Active and passive attacks types

Attacks

Protocol based Data based

Communication protocol Network protocol

Hash attack Data
exposure
Flooding
attack
Sniffing
Sniffing Denial of MaliciousVM
Worm hole attack
attack service creation
Pre-shared attack
Selective
key attack SSL forward
stripping attack

Figure 1.10 Classification of IoT attack types [32]

1.4 Challenges and future directions

IoMT-enabled healthcare applications are at fast and implementation of these

connected devices will help one to improve healthcare sector. Although automation
of health services is increasing, there are few challenges that need to be addressed.
1. Insecure data transfer among the connected devices and data theft will be a
severe privacy issue with the patient’s data.
2. Communication protocols and the risk associated with them is a challenging one.
3. Integration of IoMT with other emerging technologies is also a challenging task.
4. Data analytics and cloud computing are the challenges associated with IoMT.
5. Interoperability within IoMT is another challenging task.
6. Unstructured data and the data growth at exponential rate, memory of the
system are the challenging tasks in medical domain.
7. Hardware, software implementations and design optimization are also impor-
tant challenges for IoMT.
12 The Internet of medical things

8. Accuracy and precision are also two critical design factors for IoMT.
Inaccurate results could be harmful to the patients.
Standard protocols, 5G-enabled emerging technologies like edge computing,
explainable AI, can be integrated with IoMT for an effective use of IoMT in
healthcare sector.

1.5 Conclusion
Smart healthcare is a current and well-researched area. IoMT, AI, cloud computing,
and wireless sensor networks are all part of the smart healthcare. The main outcome
of this chapter is to highlight the recent advances and applications of IoMT in the
healthcare sector. The potential use of IoMT in handling the COVID-19 pandemic
is discussed. Security attacks and various mitigation techniques were also addres-
sed. We hope that this survey will benefit the research community in understanding
the role of IoMT in smart healthcare and to adopt this technology well in the
healthcare sector.

References
[1] L. Atzori, A. Iera and G. Morabito, The Internet of Things: A survey.
Comput. Networks 2010; 54(15): 2787–2805.
[2] J. Fenn, Gartner’s hype cycle special report for 2005, 2005. http://www.
gartner. com/resources/130100/130115/gartners_hype_c. pdf.
[3] J. Gubbi, R. Buyya, S. Marusic and M. Palaniswami, Internet of Things
(IoT): A vision, architectural elements, and future directions. Future Gener.
Comput. Syst. 2013; 29(7): 1645–1660.
[4] W. Weng, P. Chen, S. He, X. Sun and H. Peng, Smart electronic textiles.
Angew. Chem. Int. Ed. 2016; 55: 6140–6169.
[5] A.K. Raeespour and A.M. Patel, Design and evaluation of a virtual private
network architecture for collaborating specialist users. Asia Pac. J. Inf.
Technol. Multimed. 2016; 5: 13–50.
[6] R.A. Stackowiak, IoT solutions overview. In Azure Internet of Things
Revealed; Springer: Berkeley, CA, USA, 2019; pp. 29–54.
[7] R. Hassan, F. Qamar, M.K. Hasan, A.H.M. Aman and A.S. Ahmed, Internet
of Things and its applications: A comprehensive survey. Symmetry 2020; 12
(10): 1674. https://doi.org/10.3390/sym12101674.
[8] G.J. Joyia, R.M. Liaqat, A. Farooq and S. Rehman, Internet of Medical
Things (IoMT): Applications, benefits and future challenges in Healthcare
domain. J. Commun. 2017; 12(4): 240–247.
[9] J.T. John and S.J. Ramson, Energy-aware duty cycle scheduling for efficient
data collection in wireless sensor networks. IJARCET 2013; 2.
[10] S. Ramson and D.J. Moni, A case study on different wireless networking tech-
nologies for remote health care. Intell. Decis. Technol. 2016; 10(4): 353–364.
 Internet of medical things (IoMT) 13

[11] F. Al-Turjman, M.H. Nawaz and U.D. Ulusar, Intelligence in the Internet of
Medical Things era: A systematic review of current and future trends.
Comput. Commun. 2020; 150: 644–660.
[12] M. Muzammal, R. Talat, A.H. Sodhro and S. Pirbhulal, A multi-sensor data
fusion enabled ensemble approach form Medical data from body sensor
networks. Inf. Fusion 2020; 53: 155–164.
[13] T. Van Steenkiste, D. Deschrijver and T. Dhaene, “Sensor fusion using
backward shortcut connections for sleep apnea detection in multi-modal
data,” 2019, arXiv:1912.06879. [Online]. Available: http://arxiv.org/abs/
1912.06879.
[14] K. Lin, Y. Li, J. Sun, D. Zhou and Q. Zhang, Multi-sensor fusion for body
sensor network in medical human–robot interaction scenario. Inf. Fusion
2020; 57: 15–26.
[15] F. Al-Shargie, Fusion of fNIRS and EEG signals: Mental stress study.
engrXiv 2019; 2019: 1–5. doi: 10.31224/osf.io/kaqew.
[16] I. Cabria and I. Gondra, MRI segmentation fusion for brain tumor detection.
Inf. Fusion 2017; 36: 1–9.
[17] D. Fabiano and S. Canavan, Emotion recognition using fused physiological
signals. In Proc. 8th Int. Conf. Affect. Comput. Intell. Interact. (ACII),
Cambridge, UK, 2019; pp. 42–48.
[18] W. Zhang, J. Yang, H. Su, M. Kumar and Y. Mao, Medical data fusion
algorithm based on Internet of Things. Pers. Ubiquitous Comput. 2018; 22
(5–6): 895–902.
[19] J. Du, W. Li and H. Tan, Intrinsic image decomposition-based grey and
pseudo-color medical image fusion. IEEE Access 2019; 7: 56443–56456.
[20] C. Chen, R. Jafari and N. Kehtarnavaz, A real-time human action recogni-
tion system using depth and inertial sensor fusion. IEEE Sens. J. 2016; 16(3):
773–781.
[21] Z. Baloch, F.K. Shaikh and M.A. Unar, A context-aware data fusion
approach for health-IoT. Int. J. Inf. Technol. 2018; 10(3): 241–245.
[22] I. Ud Din, A. Almogren, M. Guizani and M. Zuair, A decade of Internet of
Things: Analysis in the light of healthcare applications. IEEE Access 2019;
7: 89967–89979. doi: 10.1109/ACCESS.2019.2927082.
[23] D.S.W. Ting, L. Carin, V. Dzau and T.Y. Wong, Digital technology and
COVID-19. Nat. Med. 2020; 26: 7.
[24] Y. Song, J. Jiang, X. Wang, D. Yang and C. Bai, Prospect and application of
Internet of Things technology for prevention of SARIs. Clin. eHealth 2020;
3: 1–4. doi: 10.1016/j.ceh.2020.02.001.
[25] D. Koh. 2020. Temp Pal Smart Thermometer Helps Reduce COVID-
19 Spread in Hospitals. https://www.mobihealthnews.com/news/asia-paci-
fic/temp-pal-smart-thermometer-helps-reduce-covid-19-spread-hospitals.
[26] Microsoft. 2020. Trying to Shield Hospital Staff and Patients from COVID-19
with Help from AI, Cloud, and Intelligent Edge – Asia News Center. https://
news.microsoft.com/apac/2020/03/31/trying-to-shield-hospital-staff-and-
patients-from-covid-19-with-help-from-ai-cloud-and-intelligent-edge.
14 The Internet of medical things

[27] S. Swayamsiddha and C. Mohanty, Application of cognitive Internet of

Medical Things for COVID-19 pandemic. Diabetes Metab. Syndr. 2020; 14
(5): 911–915. https://doi.org/10.1016/j.dsx.2020.06.014.
[28] W. Ejaz and M. Ibnkahla, Multiband spectrum sensing and resource allocation
for IoT in cognitive 5G networks. IEEE Internet Things J. 2017; 5(1): 150e63.
[29] S. Swayamsiddha and C. Mohanty, Application of cognitive Internet of
Medical Things for COVID-19 pandemic. Diabetes Metab. Syndr. 2020;
14(5): 911–915.
[30] R. Pratap Singh, M. Javaid, A. Haleem, R. Vaishya and S. Ali, Internet of
Medical Things (IoMT) for orthopaedic in COVID-19 pandemic: Roles,
challenges, and applications. J. Clin. Orthop. Trauma 2020; 11(4): 713–717.
doi: 10.1016/j.jcot.2020.05.011. Epub 2020 May 15. Erratum in: J Clin
Orthop Trauma. 2020 Nov-Dec;11(6):1169–1171. PMID: 32425428;
PMCID: PMC7227564.
[31] S. Bhatt and P.R. Ragiri, Security trends in Internet of Things: A survey. SN
Appl. Sci. 2021; 3(1): 1–14.
[32] A. Abdul-Ghani Hezam, D. Konstantas and M. Mahyoub, A comprehensive
IoT attacks survey based on a building-blocked reference model. Int. J. Adv.
Comput. Sci. Appl. 2018; 9: 355–373.
[33] C.-H. Huang and K.-W. Cheng, RFID technology combined with IoT
application in medical nursing system. Bull. Networking Comput. Syst.
Software 2014; 3(1): 20–24. ISSN: 21865140.
[34] Y.J. Fan, Y.H. Yin, L.D. Xu, Y. Zeng and F. Wu, IoT-based smart rehabi-
litation system. IEEE Trans. Ind. Inf. 2014; 10(2). EISSN: 2395–0072.
[35] K. Divya Krishna, V. Akkala, R. Bharath, et al., Computer aided abnormality
detection for kidney on FPGA based IoT enabled portable ultrasound imaging
system. IRBM 2016; 37: 189–197. 1959-0318. doi: 10.1016/j.irbm.2016.05.001.
[36] G. Matar, J.M. Lina, J. Carrier, A. Riley and G. Kaddoum, Internet of Things
in sleep monitoring: An application for posture recognition using supervised
learning. In 2016 IEEE 18th International Conference on e-health
Networking, Applications and Services (Healthcom), IEEE, 2016; pp. 1–6.
[37] S.S. Al-Majeed, I.S. Al-Mejibli and J. Karam, Home telehealth by Internet of
Things (IoT). In Canadian Conference on Electrical and Computer
Engineering Halifax, Canada, 3–6, 2015. doi: 978-1-4799-5829-0.
[38] I. Chiuchisan and O. Geman, An approach of a decision support and home
monitoring system for patients with neurological disorders using Internet of
Things concepts. WSEAS Trans. Syst. 2014; 13. E-ISSN: 22242678.
[39] K.B. Sundhara Kumar and K. Bairavi, IoT based health monitoring system
for autistic patients. In Symposium on Big Data and Cloud Computing
Challenges (ISBCC – 16’), Smart Innovation, Systems and Technologies 49,
2016, ’ Springer International Publishing, Switzerland, 2016. doi: 10.1007/
978-3-31930348-2-32.
[40] K. Motwani, D. Mirchandani, Y. Rohra, H. Tarachandani and A. Yeole,
Smart nursing home patient monitoring system. Imp. J. Interdiscip. Res.
(IJIR) 2016; 2(6). ISSN: 2454-1362.
 Internet of medical things (IoMT) 15

[41] P. Chavan, P. More, N. Thorat, S. Yewale and P. Dhade, ECG – Remote

patient monitoring using cloud computing. Imp. J. Interdiscip. Res. (IJIR)
2016; 2(2). ISSN: 2454-1362.
[42] A.S. Yeole and D.R. Kalbande, Use of Internet of Things (IoT) in healthcare:
A survey. In Proceedings of the ACM Symposium on Women in Research
2016, 2016; pp. 71–76.
[43] H. Arbat, S. Choudhary and K. Bala, IOT smart health band. Imp. J.
Interdiscip. Res. (IJIR) 2016; 2(5). ISSN: 2454-1362.
[44] L. Yu, Y. Lu and X.J. Zhu, Smart hospital based on Internet of Things.
J. Networks 2012; 7(10). doi: 10.43.4/jnw.7.10.1654-1661.
[45] K.M. Chaman Kumar, A new methodology for monitoring OSA patients
based on IoT. Int. J. Innovative Res. Dev. 2016; 5(2). ISSN: 2278-0211.
[46] R. Singh, A proposal for mobile E-care health service system using IOT for
Indian scenario. J. Network Commun. Emerg. Technol. (JNCET) 2016; 6(1).
ISSN: 2395-5317.
[47] C. Puri, A. Ukil and S. Bandyopadhyay, iCarMa: Inexpensive cardiac
arrhythmia management – An IoT healthcare analytics solution. In IoT of
Health’16, 2016. doi: 10.1145/2933566.2933567.
[48] V. Chandel, A. Sinharay and N. Ahmed, Exploiting IMU sensors for IOT enabled
health monitoring. In IoT of Health’16, 2016. doi: 10.1145/2933566.2933569.
[49] M. Fischer and M. Lam, From books to bots: Using medical literature to create a
chat bot. In IoT of Health’16, Singapore, 2016. doi: 10.1145/2933566.2933573.
[50] A. Ghose, P. Sinha, C. Bhaumik, A. Sinha, A. Agrawal and A.D. Choudhury,
UbiHeld – Ubiquitous healthcare monitoring system for elderly and chronic
patients. In UbiComp’13, Zurich, Switzerland, September 8–12, 2013. doi:
10.1145/2494091.2497331.
[51] H. Li, H. Wang, W. Yin, Y. Li, Y. Qian and F. Hu, Development of a remote
monitoring system for henhouse environment based on IoT technology.
Future Internet 2015; 7: 329–341.
[52] N.S. Kim, K. Lee and J.H. Ryu, Study on IoT based wild vegetation com-
munity ecological monitoring system. In Proceedings of the 2015 Seventh
International Conference on Ubiquitous and Future Networks, Sapporo,
Japan, 7–10, 2015; pp. 311–316.
[53] R. Nordin, H. Mohamad, M. Behjati, et al., The world-first deployment of
narrowband IoT for rural hydrological monitoring in UNESCO biosphere
environment. In Proceedings of the 2017 IEEE 4th International Conference
on Smart Instrumentation, Measurement and Application (ICSIMA),
Putrajaya, Malaysia, 28–30, 2017; pp. 1–5.
[54] M.C. Yuen, S.Y. Chu, W. Hong Chu, H. Shuen Cheng, H. Lam Ng and S.
Pang Yuen, A low-cost IoT smart home system. Int. J. Eng. Technol. 2018;
7: 3143–3147.
[55] D.W. Sukmaningsih, W. Suparta, A. Trisetyarso, B.S. Abbas and C.H. Kang,
Proposing smart disaster management in urban area. In Proceedings of the
Asian Conference on Intelligent Information and Database Systems,
Yogyakarta, Indonesia, 8–11, 2019; pp. 3–16.
16 The Internet of medical things

[56] F. Ahamad, M. Latif, M. Yusoff, M. Khan and L. Juneng, So near yet so

different: Surface ozone at three sites in Malaysia. EES 2019; 228: 012024.
[CrossRef].
[57] W. Suparta, K.M. Alhasa and M.S.J. Singh, Preliminary development of
greenhouse gases system data logger using microcontroller Netduino. Adv.
Sci. Lett. 2017; 23: 1398–1402.
[58] F. Montori, L. Bedogni and L. Bononi, A collaborative Internet of Things
architecture for smart cities and environmental monitoring. IEEE Internet
Things J. 2017; 5: 592–605.
[59] P. Zia, T. Liu and W. Han, Application specific digital forensics investiga-
tive model in Internet of Things (IoT). In Proceedings of the 12th
International Conference on Availability, Reliability and Security, Reggio
Calabria, Italy, 2017; pp. 1–7.
[60] Y.B. Lin, Y.W. Lin, C.Y. Hsiao and S.Y. Wang, Location-based IoT appli-
cations on campus: The IoT talk approach. Pervasive Mob. Comput. 2017;
40: 660–673.
[61] X. Zeng, S.K. Garg, P. Strazdins, P.P. Jayaraman, D. Georgakopoulos and R.
Ranjan, IOTSim: A simulator for analysing IoT applications. J. Syst. Archit.
2017; 72: 93–107.
[62] M.A. Jabbar and R. Aluvalu, Smart cities in India: Are we smart enough? In
2017 International Conference On Smart Technologies For Smart Nation
(SmartTechCon), 2017; pp. 1023–1026. doi: 10.1109/SmartTechCon.
2017.8358525.
[63] S. Chen, B. Liu, X. Chen, Y. Zhang and G. Huang, Framework for
adaptive computation offloading in IoT applications. In Proceedings of
the 9th Asia-Pacific Symposium on Internetware, Shanghai, China, 23,
2017; pp. 1–6.
[64] A. Urbieta, A. González-Beltrán, S.B. Mokhtar, M.A. Hossain and L. Capra,
Adaptive and context-aware service composition for IoT-based smart cities.
Future Gener. Comput. Syst. 2017; 76: 262–274. [CrossRef].
[65] D. Seo, Y.B. Jeon, S.H. Lee and K.H. Lee, Cloud computing for ubiquitous
computing on M2M and IoT environment mobile application. Cluster
Comput. 2016; 19: 1001–1013.
[66] C. Lee, C. Wang, E. Kim and S. Helal, Blueprint flow: A declarative service
composition framework for cloud applications. IEEE Access 2017; 5:
17634–17643.
[67] A. Akbar, G. Kousiouris, H. Pervaiz, et al., Real-time probabilistic data
fusion for large-scale IoT applications. IEEE Access 2018; 6: 10015–10027.
[68] X. Sun and N. Ansari, Traffic load balancing among brokers at the IoT
application layer. IEEE Trans. Netw. Serv. Manage. 2017; 15: 489–502.
[69] X. Sun and N. Ansari, Dynamic resource caching in the IoT application layer
for smart cities. IEEE Internet Things J. 2017; 5: 606–613.
[70] G.G. Krishna, G. Krishna and N. Bhalaji, Analysis of routing protocol for
low-power and lossy networks in IoT real time applications. Procedia
Comput. Sci. 2016; 87: 270–274.
 Internet of medical things (IoMT) 17

[71] P.G.V. Naranjo, Z. Pooranian, M. Shojafar, M. Conti and R. Buyya,

FOCAN: A Fog-supported smart city network architecture for management
of applications in the Internet of Everything environments. J. Parallel
Distrib. Comput. 2019; 132: 274–283.
[72] S. Kim and S. Kim, User preference for an IoT healthcare application for
lifestyle disease management. Telecommun. Policy 2018; 42: 304–314.
[73] X. Fafoutis, L. Clare, N. Grabham, et al., Energy neutral activity monitoring:
Wearables powered by smart inductive charging surfaces. In Proceedings of
the 2016 13th Annual IEEE International Conference on Sensing,
Communication, and Networking (SECON), London, UK, 27–30, 2006;
pp. 1–9.
[74] F. Jimenez and R. Torres, Building an IoT-aware healthcare monitoring sys-
tem. In Proceedings of the 2015 34th International Conference of the Chilean
Computer Science Society (SCCC), Santiago, Chile, 9–13, 2015; pp. 1–4.
[75] Y. Ding, S. Gang and J. Hong, The design of home monitoring system by
remote mobile medical. In Proceedings of the 2015 7th International
Conference on Information Technology in Medicine and Education (ITME),
Huangshan, China, 13–15, 2015; pp. 278–281.
[76] H.F. Atlam and G.B. Wills, IoT security, privacy, safety and ethics. In
Digital Twin Technologies and Smart Cities; Springer: Cham, Switzerland,
2020; pp. 123–149.
[77] Z. Baloch, F.K. Shaikh and M.A. Unar, A context-aware data fusion
approach for health-IoT. Int. J. Inf. Technol. 2018; 10: 241–245.
[78] V. Subrahmanyam, M.A. Zubair, A. Kumar and P. Rajalakshmi, A low
power minimal error IEEE 802.15. 4 Transceiver for heart monitoring in IoT
applications. Wirel. Pers. Commun. 2018; 100: 611–629.
[79] S.C. Lin, C.Y. Wen and W.A. Sethares, Two-tier device-based authentica-
tion protocol against PUEA attacks for IoT applications. IEEE Trans. Signal
Inf. Process. Over Networks 2017; 4: 33–47.
[80] H.A. Damis, N. Khalid, R. Mirzavand, H.J. Chung and P. Mousavi,
Investigation of epidermal loop antennas for biotelemetry IoT applications.
IEEE Access 2018; 6: 15806–15815.
[81] M. Elappila, S. Chinara and D.R. Parhi, Survivable path routing in WSN for
IoT applications. Pervasive Mob. Comput. 2018; 43: 49–63.
[82] J. Jebadurai and J.D. Peter, Super-resolution of retinal images using multi-
kernel SVR for IoT healthcare applications. Future Gener. Comput. Syst.
2018; 83: 338–346.
[83] H. Malik, M.M. Alam, Y. Le Moullec and A. Kuusik, Narrow band-IoT
performance analysis for healthcare applications. Procedia Comput. Sci.
2018; 130: 1077–1083.
[84] R. Hamdan, Human factors for IoT services utilization for health informa-
tion exchange. J. Theor. Appl. Inf. Technol. 2018; 96: 2095–2105.
[85] M. Shahidul Islam, M.T. Islam, A.F. Almutairi, G.K. Beng, N. Misran and
N. Amin, Monitoring of the human body signal through the Internet of
Things (IoT) based LoRa wireless network system. Appl. Sci. 2019; 9: 1884.
18 The Internet of medical things

[86] M.A. Dauwed, J. Yahaya, Z. Mansor and A.R. Hamdan, Determinants of

Internet of Things services utilization in health information exchange.
J. Eng. Appl. Sci. 2018; 13: 10490–10501.
[87] M.A. Jabbar, S. Samreen and R. Aluvalu, The future of health care: Machine
learning. Int. J. Eng. Technol. 2018; 7(4): 23–25.
[88] I. Chiuchisan, H.-N. Costin and O. Geman, Adopting the Internet of Things
technologies in health care systems. In Proc. Int. Conf. Expo. Electr. Power
Eng. (EPE), Iasi, Romania, 2014, pp. 532–535.
[89] A. AwadAbdellatif, A. Emam, C.-F. Chiasserini, A. Mohamed, A. Jaoua and
R. Ward, Edge-based compression and classification for smart healthcare
systems: Concept, implementation and evaluation. Expert Syst. Appl. 2019;
117: 1–14.
[90] G. Muhammad, M.F. Alhamid, M. Alsulaiman and B. Gupta, Edge com-
puting with cloud for voice disorder assessment and treatment. IEEE
Commun. Mag. 2018; 56(4): 60–65.
[91] M.Z. Uddin, A wearable sensor-based activity prediction system to facilitate
edge computing in smart health care system. J. Parallel Distrib. Comput.
2019; 123: 46–53.
[92] A. Al-nasheri, G. Muhammad, M. Alsulaiman and Z. Ali, Investigation of
voice pathology detection and classification on different frequency regions
using correlation functions. J. Voice 2017; 31(1): 3–15.
[93] J.C. Wang, C.Y. Ng and R.H. Brook, Response to COVID-19 in Taiwan: Big
data analytics, new technology, and proactive testing. J. Am. Med. Assoc.
2020; 323(14): 2.
[94] Y. Lee. 2020. Covid-19: Taiwan’s New ‘electronic Fence’ for Quarantines
Leads Wave of Virus Monitoring. https://www.thestar.com.my/tech/tech-
news/2020/03/20/covid-19-taiwans-new-electronic-fence-for-quarantines-
leads-wave-of-virus-monitoring.
[95] H. Leite, T. Gruber and I.R. Hodgkinson, Flattening the infection curve –
Understanding the role of telehealth in managing COVID-19. Leader.
Health Serv. 2019; 33(2): 1751–1879. doi: 10.1108/LHS-05-2020-084.
[96] L. Ferretti, Quantifying dynamics of SARS-CoV-2 transmission suggests
that epidemic control with digital contact tracing. Science 2020; 31. doi:
10.1126/science.abb6936. Mar 2020.
[97] Bosch. 2020. Combating the Coronavirus Pandemic: Bosch Develops Rapid
Test for COVID-19. https://www.bosch.com/stories/vivalytic-rapid-test-for-
covid-19.
 Chapter 2
Non-invasive psycho-physiological driver
monitoring through IoT-oriented systems
Luca Davoli1, Veronica Mattioli2, Sara Gambetta3, Laura
Belli1, Luca Carnevali3, Marco Martalò4*, Andrea Sgoifo3,
Riccardo Raheli2 and Gianluigi Ferrari1

The definition, analysis, and implementation of in-vehicle monitoring systems that

collect data which are informative of the status of the joint driver-vehicle system
represent a topic of strong interest from both academic players and industrial
manufacturers. Many external factors, such as road design, road layout, traffic flow,
and weather, can influence and increase driving-related stress, potentially increas-
ing risks. The ubiquitous diffusion of Internet of Things (IoT) technologies allows
one to collect heterogeneous data that can build the foundation for driver’s psy-
chophysiological characterization, with the aim of improving safety and security
while driving. This chapter evaluates and discusses the feasibility and usefulness of
a noninvasive IoT-oriented driver monitoring infrastructure aiming at collecting
physiological parameters (such as heart rate variability, HRV) that may be adopted
as biomarkers of the driver’s psychophysiological state in different driving
scenarios.

2.1 Introduction
Driving is one of the major experiences linking together people: every day, while
travelling by car, drivers are exposed to different events and situations, which can
impact on their psychophysiological state. External factors such as road design

1
Internet of Things (IoT) Lab, Department of Engineering and Architecture, University of Parma, Parma,
Italy
2
Multimedia Lab, Department of Engineering and Architecture, University of Parma, Parma, Italy
3
Stress Physiology Lab, Department of Chemistry, Life Sciences and Environmental Sustainability,
University of Parma, Parma, Italy
4
Networks for Humans (Net4U) Lab, Department of Electrical and Electronic Engineering, University of
Cagliari, Cagliari, Italy
*
The work of Marco Martalò was carried out at the following place: in 2020, at the IoT Lab of the
University of Parma; and in 2021 while collaborating with the IoT Lab.
20 The Internet of medical things

(motorways vs. rural roads vs. city roads, etc.), road layout (straight vs. curves,
steep road vs. downhill road, etc.), traffic flow (high vs. low), and weather can
influence and increase driving-related stress. To this end, literature studies exam-
ined the relationship between traffic conditions and stress levels and, as expected,
they found that driving-related stress is greater in high traffic jam areas rather than
in low congestion ones [1–3]. Moreover, it has been observed that these stressful
situations may alter the driver’s psychophysiological state.
In this regard, HRV represents an indirect and noninvasive measurement of
beat-to-beat temporal changes in heart rate (HR), which reflects cardiac autonomic
influences, particularly of vagal origin, at the sinoatrial node of the heart. HRV
analysis has been extensively applied in various research fields, including psy-
chophysiology, cardiology, and psychiatry and has been increasingly recognized as
a biomarker of health and stress. In fact, a healthy subject is characterized by higher
levels of resting HRV that are in turn associated with better flexibility and adapt-
ability to environmental challenges [4]. Moreover, HRV analysis is able to index
psychophysiological states during stressful conditions or mental efforts (such as
during driving activities) [4]. Indeed, lower tonic HRV has been associated with
psychosocial stress and mental workload [5–7], whereas higher HRV reflects the
capability of an individual to successfully adapt to external stimuli and manifesting
itself in a reduction in daily performance [8]. An example of the potential utility of
HRV as a biomarker of physiological driver’s state is shown in [9], where driving-
related stress is associated with a reduction in HRV indexes of vagal tone.
From a more technological perspective, the definition and analysis of mon-
itoring systems inside vehicles that are able to simultaneously collect physiolo-
gical indexes and useful data to determine the status of the driver–vehicle system
is a topic of strong interest. The final aim for both academic players and industrial
manufacturer is to improve safety and security while driving for both drivers and
passengers [10]. Furthermore, recent technological progress has introduced new
possibilities with respect to traditional manual driving, especially with regard to
mechanisms aiming at vehicle—e.g., cars, trucks, etc.—driving assistance, such
as onboard advanced driver-assistance systems (ADAS) [11]. In detail, examples
of ADAS mechanisms equipping modern vehicles are antilock braking system
[12], adaptive cruise control [13], electronic stability control (ESC) [14], lane
departure warning system [15], forward collision warnings [16], traffic sign
recognition [17], automotive night vision [18], alcohol ignition interlock devices
[19], collision avoidance systems [20], and driver drowsiness detection [21].
Then, as a common objective, all these ADAS aim to compensate for human
errors, in order to reduce road fatalities—activating alarms or, when necessary,
taking control of the vehicle itself.
Furthermore, focusing on a more general perspective, in the last years the
ubiquitous diffusion of the IoT paradigm [22] has influenced and changed our
lifestyle, thanks to the ability to interconnect heterogeneous devices (such as sen-
sors, actuators, or more in general, smart objects [23,24]), combining different
technologies and communication protocols to build services for end users [25].
Thanks to this heterogeneity, IoT applications are innumerable—nowadays often
 Noninvasive psychophysiological driver monitoring 21

requiring to outsource some processing efforts to cloud-based infrastructures

[26]—and, among different possible scenarios, one of the most relevant ones is
linked to the automotive industry, with the aim of acquiring a better knowledge of
the driver–vehicle system as a whole.
Then, looking at the interaction level of all these paradigms, the majority of
existing ADAS do not take into account aspects related to the psychophysiological
state of the driver. This would require continuous monitoring and understanding of
the driver’s physical, emotional, and physiological state, an effective communica-
tion of the ADAS decisions to the driver, or a direct action on the vehicle. Although
a monitoring system with these characteristics is very challenging to obtain, ADAS
provided with information about the driver’s psychophysiological state could take
more contextualized actions, implementing complex decisions that are compatible
with the driver’s possible reactions. Another aspect that should be considered in the
field of in-vehicle monitoring systems is the degree of intrusiveness of the sensors
employed for collecting driver’s physiological parameters, which in fact should not
interfere—or minimize as much as possible their interference—with the driving
activity and therefore should be selected accordingly.
Due to these premises, the aim of this chapter is 2-fold: (i) propose a non-
invasive IoT driver monitoring infrastructure, aiming at collecting data useful to
estimate the psychophysiological status of the driver, as a base for subsequent
actions, and (ii) evaluate the usefulness of HRV parameters as biomarkers of a
driver’s psychophysiological state in different driving scenarios.
The rest of the chapter is organized as follows. A review of heterogeneous
technologies for human driver monitoring is presented in Section 2.2. Section 2.3
describes the in-vehicle IoT monitoring system and the experimental setup. In
Section 2.4, a preliminary evaluation of collected data is presented. Finally, some
conclusions and propose future research directions are drawn in Section 2.5.

2.2 Heterogeneous driver monitoring

When considering technologies for human driver monitoring activities, a focal
point that should be taken into account is the extent to which the end user (i.e., the
driver) takes up these technologies. This is strictly dependent on their unobtru-
siveness, i.e., their capacity to be perceived by the user as not infringing upon his/
her privacy or interfering with his/her driving activity. Taking these aspects into
account, IoT sensing devices can be adopted as monitoring technologies and could
be selected on the basis of their mobility degree. Mobile sensors (e.g., wearable and
sensors-equipped smartphones) have become practical and appealing—thanks to
the miniaturization of the components—and could also be exploited for monitoring
driver’s behavior and physiology [27]. On the other hand, stationary sensors are
typically installed in the environment that needs to be monitored (e.g., in the
vehicle’s cabin) and act without any physical contact with the user, thus being well
suited for unobtrusive monitoring—an example of a stationary sensor may be a
video and/or thermal camera positioned in the vehicle’s cabin.
22 The Internet of medical things

2.2.1 Wearable and inertial sensors

One of the uprising unobtrusive ways to monitor individual physiological para-
meters while driving is through the collection and analysis of biological signals.
For example, the electrocardiogram (ECG) gives uniqueness, universality, and
permanence [28] in the measurement, thanks to its working principle by which the
biometric signal is the result of the electrical conduction through the heart needed
for its contraction [29]. Then, ECG analysis directly inside the vehicle’s cabin may
be useful for assessing parameters (e.g., HRV) related to both mental and physical
stress, and workload [30–32] for automatic vehicle settings customization (e.g.,
biometric authentication for ignition lock [33]). Moreover, the respiratory signal
may be exploited in order to obtain additional physiological indicators (e.g., the
respiratory rate, RR) that are useful in the automotive scenarios. In particular, these
data could be informative of the driver’s workload and stress status. Nevertheless, it
could be interesting to investigate how the combination of these measurements with
vision data (either by normal or thermal cameras) would allow one to estimate a
classification index for workload or stress level of the driver [34]. Additional
mobile-sensing elements (often worn by the person to be monitored) are inertial
measurement units (IMUs), which allow the estimation of the driver motion level
through three-axis internal accelerometers and gyroscopes. More in detail, IMUs
allow both motion and attitude/pose estimations, but they require to be carefully
chosen and calibrated—because of bias instability, scaling, alignment, and tem-
perature dependence. All these issues should be carefully considered especially in
operating areas that require limitations in IMU’s size, weight, and cost limitations.
With particular regard to the automotive field, inertial sensing mechanisms can
intervene in different scenarios, from driver’s movements monitoring when seated
in the vehicle’s cabin to safety areas—like vehicle dynamics control (e.g., ESC)
and passenger restraint systems (e.g., seat-belt pretensioner and airbags).

2.2.2 Camera sensors

With regards to unobtrusive and stationary sensing technologies to be used in driver
monitoring contexts, a particularly interesting one can be found in imaging-based
solutions, such as video cameras. Unfortunately, as for sensing mechanisms discussed
in Section 2.2.1, even video cameras are not exempt from drawbacks, especially
depending on swinging lighting and atmospheric conditions—which may interfere
with driver’s movements monitoring and physiological signal recognition. Considering
these aspects, a way to overcome this limitation is the deployment of thermal cameras,
being more robust to adverse light conditions compared to canonical imaging sensors.
More in detail, long wavelength infrared cameras [35] can detect objects in different
environmental conditions (e.g., rain, darkness, in the presence of fog, etc.), are unaf-
fected by sun glare (improving situational awareness [36]), and are more robust against
reflections, shadows, and car headlights [37]. Even in this case, the price to be paid is
that the thermal camera’s video resolution is typically lower than that of traditional
video sensors. Nevertheless, more recent thermal cameras are able to produce high-
quality video streams. Finally, in addition to the driver’s monitoring based on thermal
 Noninvasive psychophysiological driver monitoring 23

cameras, useful integration could involve the use of normal and general-purpose ima-
ging sensors, such as in-vehicle fixed cameras, or the front camera of a smartphone
[38,39]. In detail, these additional video sensing elements can integrate the experi-
mental data obtained by thermal cameras, thanks to their well-known portability, thus
representing an optimal way for gathering information [40].

2.3 In-vehicle IoT-oriented monitoring architecture

2.3.1 Experimental setup
Given the different monitoring entities discussed in Section 2.2, the noninvasive IoT-
oriented driver monitoring architecture proposed in this chapter is shown in
Figure 2.1. In order to collect the driver’s physiological data, an Equivital
EQ02 LifeMonitor [41] sensor belt has been adopted. This wearable sensing device is
composed by two elements: (i) a chest belt containing fabric electrodes placed in
good contact with the driver’s skin, and (ii) a sensor electronics module (SEM) col-
lecting ECG and respiratory signals, skin temperature, and three-axis accelerometer
data. Then, as shown in Figure 2.2, on a practical side the belt securely holds the SEM
on the driver’s body through a specific pocket on the left side of the belt itself.
Moreover, in order to collect thermal information from the body (i.e., face) of the
driver and the surrounding environment, an FLIR One Pro LT [42] thermal camera
has been included in the experimental IoT-oriented monitoring architecture. In detail,
this video-capturing device is intended to work as an external “dongle” to be plugged
into a smartphone (running either Android or iOS as operating system) and subse-
quently positioned and installed inside the vehicle, as shown in Figure 2.3. To this

Data processing

V2X BLE

ECU
HUB

Wi-Fi
CAN

Figure 2.1 In-vehicle IoT-oriented monitoring architecture

24 The Internet of medical things

Figure 2.2 Equivital EQ02 LifeMonitor sensor belt positioning

Figure 2.3 FLIR One Pro LT thermal camera positioning inside the vehicle

end, it is necessary to carefully and accurately define the position of the smartphone
connected with the thermal camera; this is as much true as it represents a trade-off
between the quality of the framing—further image-processing and analysis tasks
require a frontal viewpoint—and the degree of obtrusiveness for the driver’s per-
spective toward the windscreen.

2.3.2 HRV analysis

The experimental testbed has been used to obtain high-quality ECG recordings under
stationary conditions in order to perform a first HRV analysis. In detail, ECG signals
are digital waves converted through specific processing tasks and, aiming at obtain-
ing a good time resolution, a sampling rate of at least 250 Hz is recommended. Then,
 Noninvasive psychophysiological driver monitoring 25

HRV is quantified by analyzing the variations of time intervals between consecutive

normal heart beats. To this end, an inter-beat interval may be defined as the time
between consecutive R-waves peaks of the ECG (R–R interval), while the time-
course of the R–R interval is called tachogram and further quantitative analysis of
this curve will allow the estimation of HRV parameters. HRV analysis is performed
by applying time-domain and frequency-domain methods.
Time-domain parameters are calculated with mathematical approaches to
measure the amount of variability present in a specific time period in a continuous
ECG signal. The most frequent time domain indexes adopted for HRV analysis are
detailed in Table 2.1: (i) standard deviation of the R–R intervals (SDNN), (ii) root-
mean square of successive differences between adjacent R–R intervals (RMSSD),
and (iii) percentage of successive R–R interval differences exceeding 50 ms
(pNN50). In detail, the SDNN estimates overall HRV and includes the contribution
of both branches of the autonomic nervous system (ANS) to HR variations. The
RMSSD estimates vagally mediated changes in HR [43]. Finally, the pNN50
quantifies the percentage of successive R–R interval differences that are larger than
50 ms and reflects the vagal tone [43].
Instead, frequency-domain analysis requires to filter the signal into different
bands. In fact, the power spectral analysis decomposes a time-dependent fluc-
tuating signal into its sinusoidal components, allowing one to detect and quantify
the amount of cyclical variation present at different frequencies [44]. Moreover,
it provides information on how the power is distributed as a function of fre-
quency. Thus, in a typical power spectral density curve, three main frequency
bands can be identified, as detailed in Table 2.2: (i) very low frequency (VLF),

Table 2.1 HRV indexes considered in the time domain

Variable Description Physiological origin

SDNN Standard deviation of the R–R intervals Cyclic components responsible
for heart rate variability
RMSSD Root-mean square of successive differences Vagal tone
between adjacent R–R intervals
pNN50 Percentage of successive R–R interval dif- Vagal tone
ferences exceeding 50 ms

Table 2.2 HRV indexes considered in the frequency domain

Variable Description Physiological origin

VLF Very low frequency (<0.04 Hz) Long-term regulation mechanisms,
thermoregulation and hormonal me-
chanism
LF Low frequency (0.04–0.15 Hz) Mix of sympathetic and vagal activity
HF High frequency (0.15–0.4 Hz) Vagal tone
26 The Internet of medical things

(ii) low frequency (LF), and (iii) high frequency (HF) bands. The VLF compo-
nent reflects R–R interval variations that are due to long-term regulation
mechanisms (e.g., thermoregulation and hormonal mechanisms). The LF band
reflects a mix between sympathetic and vagal influences. The HF band reflects
vagal tone and is linked to respiratory-related changes in cardiac autonomic
modulation [45].

2.4 Experimental performance evaluation

2.4.1 Operating protocol for data collection
In order to collect information on the driver’s psychophysiological state in different
driving scenarios, driving tests should be performed according to a well-defined
operating protocol. Both smooth and fast driving should be included in the analysis
to evaluate the driver’s response to different external stimuli, associated with dif-
ferent amount of perceived stress. All these tests should then take place on both
urban and highway roads in situations of smooth and heavy traffic.
The experimental driver monitoring has been conducted on a 26-year-old
Italian female subject, separating the driving protocol into six time intervals, as
shown in Figure 2.4: (i) baseline, (ii) first city driving, (iii) first highway driving,
(iv) second city driving, (v) second highway driving, and (vi) rest phase. Each
driving period had a duration of at least 5 min, in the end resulting in a total driving
time of 30 min. In order to collect driver’s physiological data, the Equivital
EQ02 LifeMonitor sensor was worn by the subject throughout the driving test and
the FLIR One Pro LT thermal camera was used to simultaneously record
temperature-related information.
The overall route covered during the experimental campaign inside the city of
Parma, Italy, is shown in Figure 2.5. During the baseline period, the vehicle’s
engine was off and the subject was asked to sit still inside the vehicle to acquire
baseline data. During city and highway driving phases, urban and ring roads were
covered with an average speed of 50 and 90 km/h, respectively. Finally, at the end
of the route, data recording continued for at least 5 min during recovery conditions,
with the driver inside the car.

Baseline City 1 Highway City 2 Highway Rest

0 5 10 15 20 25 30
(min)

Recording of ECG, respiratory and thermal signals

Figure 2.4 Operating driving protocol adopted in the experimental driver

monitoring
 Noninvasive psychophysiological driver monitoring 27

City 2

City 1

Highway
Highway

Rest

Baseline

Figure 2.5 Urban and highway roads traveled in the city of Parma, Italy, during
the experimental driver monitoring

2.4.2 HR and HRV data

ECG and respiratory signals were exported from the wearable sensor by means of
its software manager, denoted as Equivital Manager. Then, raw ECG and respira-
tory data obtained from the EQ02 LifeMonitor equipment were amplified, digi-
tized, and analyzed by means of the LabChart Pro 5.0 software [46]. More in detail,
each raw ECG signal was manually inspected to ensure that all R-waves were
correctly detected and to exclude artifacts before further analysis. Then, for each
recording period, ECG signals were split in 5-min epochs and, for each epoch, HR
has been calculated by plotting the number of R-waves per time unit (dim: beats per
minute [bpm]), together with the estimation of time- and frequency-domain HRV
indexes (namely, RMSSD and HF). Finally, as for the ECG signals, for each
recording period, respiratory signals were split in 5-min epochs, and R–R was
calculated for each epoch.

2.4.3 Experimental results

In Figure 2.6, average HR temporal dynamics (dim: [bpm]) are shown. In detail, at
the beginning of the recording session, the subject showed an HR value equal to
85 bpm. Then, the HR increased during the first city driving phase and remained
constant throughout the rest of the drive. Finally, at the end of the driving session,
in the rest phase, the HR returned to baseline levels.
RMSSD values (dim: [ms]) are shown in Figure 2.7. At the beginning of the
driving session, the average RMSSD was equal to 34 ms. Then, the RMSSD
decreased during both city and highway driving phases, till increasing back to
baseline levels during the rest phase, similar to what was observed for HR.
28 The Internet of medical things

95

90
HR (bpm)

85

80
ine

y1

y2

st
wa

wa

Re
sel

Cit

Cit
gh

gh
Ba

Hi

Hi
Condition/Location

Figure 2.6 Experimental HR values obtained during the driving route

38
36
34
RMSSD (ms)

32
30
28
26
24
ine

y1

y2

st
wa

wa

Re
sel

Cit

Cit
gh

gh
Ba

Hi

Hi

Condition/Location

Figure 2.7 Experimental RMSSD values obtained during the driving route

700

600
HF (ms2)

500

400

300
ine

y1

y2

st
wa

wa

Re
sel

Cit

Cit
gh

gh
Ba

Hi

Hi

Condition/Location

Figure 2.8 Experimental HF values obtained during the driving route

In Figure 2.8, the values of HF (dim: [ms2]) are shown. As for RMSSD, the
driving phase was characterized by a reduction of HF values in both city and
highway scenarios, thus suggesting a decrease in parasympathetic modulation.
Finally, the average RR values (dim: counts per minute [cpm]) are shown in
Figure 2.9. The R–R interval was steady during the overall recording session.
 Noninvasive psychophysiological driver monitoring 29

18.5
18
17.5

RR (cpm)
17
16.5
16
15.5
15
ine

y1

y2

st
wa

wa

Re
sel

Cit

Cit
gh

gh
Ba

Hi

Hi
Condition/Location

Figure 2.9 Experimental R–R values obtained during the driving route

These preliminary results suggest that the driving experience can modulate the
psychophysiological state of the driver, as hinted by the reduction in HRV para-
meters associated with the increase of HR. Therefore, further driving recording
sessions with a larger sample size are needed to confirm these results and indicate
the extent to which HRV parameters may vary according to the psychophysiolo-
gical state of the driver in different driving scenarios. Moreover, a future integra-
tion of thermal imaging data would provide a clearer picture of the physiological
correlates of drivers’ mental effort.

2.5 Conclusions and future works

In this chapter, a preliminary noninvasive infrastructure for in-vehicle driver
monitoring of the psychophysiological status of the driver is presented. This
infrastructure is based on an IoT sensor network composed by a wearable sensor
belt, worn by the driver, and a thermal camera, installed in the vehicle’s cabin
framing the driver. Then, the protocol adopted for experimental data acquisition
campaigns, together with ECG and respiratory signals collection and analysis
during different driving scenarios, have been described. While additional experi-
mental recordings are needed to confirm these experimental results and to allow a
better investigation on the effects of driving activities on autonomic neural mod-
ulation of the cardiac function, the preliminary results on the utility of HRV para-
meters as biomarkers of a driver’s psychophysiological state are promising.
Future activities will involve the integration of thermal camera-based data and
the improvement of the IoT communication infrastructure in order to integrate
additional data sources, e.g., data collected directly from the onboard vehicular bus
and emitted from the vehicle’s electronic control unit possibly in real time.

Acknowledgments
The work of the authors is supported by the European Union’s Horizon 2020
research and innovation program ECSEL Joint Undertaking (JU) under grant
30 The Internet of medical things

agreement no. 876487, NextPerception project—“Next Generation Smart Perception

Sensors and Distributed Intelligence for Proactive Human Monitoring in Health,
Wellbeing, and Automotive Systems.” The JU received support from the European
Union’s Horizon 2020 research and innovation program and the nations involved in
the mentioned projects. The work reflects only the authors’ views; the European
Commission is not responsible for any use that may be made of the information it
contains. The work of L.D. is partially funded by the University of Parma, under
“Iniziative di Sostegno alla Ricerca di Ateneo” program, “Multi-interface IoT
sYstems for Multi-layer Information Processing” (MIoTYMIP) project.

References
[1] Hennessy DA and Wiesenthal DL. The Relationship Between Traffic
Congestion, Driver Stress and Direct Versus Indirect Coping Behaviours.
Ergonomics. 1997; 40(3): 348–361. doi:10.1080/001401397188198.
[2] Hill JD and Boyle LN. Driver Stress as Influenced by Driving Maneuvers and
Roadway Conditions. Transportation Research Part F: Traffic Psychology and
Behaviour. 2007; 10(3): 177–186. doi:10.1016/j.trf.2006.09.002.
[3] Neighbors C, Vietor NA, and Knee CR. A Motivational Model of Driving
Anger and Aggression. Personality and Social Psychology Bulletin. 2002;
28(3): 324–335. doi:10.1177/0146167202286004.
[4] Thayer JF, Hansen AL, Saus-Rose E, et al. Heart Rate Variability, Prefrontal
Neural Function, and Cognitive Performance: The Neurovisceral Integration
Perspective on Self-regulation, Adaptation, and Health. Annals of Behavioral
Medicine. 2009; 37(2): 141–153. doi:10.1007/s12160-009-9101-z.
[5] Lischke A, Jacksteit R, Mau-Moeller A, et al. Heart Rate Variability is Associated
With Psychosocial Stress in Distinct Social Domains. Journal of Psychosomatic
Research. 2018; 106: 56–61. doi:10.1016/j.jpsychores.2018.01.005.
[6] Castaldo R, Melillo P, Bracale U, et al. Acute Mental Stress Assessment via
Short Term HRV Analysis in Healthy Adults: A Systematic Review With
Meta-Analysis. Biomedical Signal Processing and Control. 2015; 18: 370–
377. doi:10.1016/j.bspc.2015.02.012.
[7] Mulder T, De Waard D, and Brookhuis KA. Estimating Mental Effort Using
Heart Rate and Heart Rate Variability. In: Stanton NA, Hedge A, Brookhuis
K, Salas E, Hendrick HW, editors. Handbook of Human Factors and
Ergonomics Methods. 1st ed. Boca Raton, FL: CRC Press; 2004. p. 227–236.
doi:10.1201/9780203489925-30.
[8] Melillo P, Bracale M, and Pecchia L. Nonlinear Heart Rate Variability
Features for Real-Life Stress Detection. Case study: Students Under Stress
Due to University Examination. BioMedical Engineering OnLine. 2011; 10
(1): 96. doi:10.1186/1475-925x-10-96.
[9] Riener A, Ferscha A, and Aly M. Heart on the Road: HRV Analysis for
Monitoring a Driver’s Affective State. In: Proceedings of the 1st International
Conference on Automotive User Interfaces and Interactive Vehicular Applications.
 Noninvasive psychophysiological driver monitoring 31

AutomotiveUI ’09. New York, NY, USA: Association for Computing Machinery;
2009. p. 99–106. doi:10.1145/1620509.1620529.
[10] Davoli L, Martalò M, Cilfone A, et al. On Driver Behavior Recognition for
Increased Safety: A Roadmap. Safety. 2020; 6(4): 1–33. doi:10.3390/
safety6040055.
[11] Ziebinski A, Cupek R, Grzechca D, et al. Review of Advanced Driver
Assistance Systems (ADAS). AIP Conference Proceedings. 2017; 1906(1):
120002. doi:10.1063/1.5012394.
[12] Satoh M and Shiraishi S. Performance of Antilock Brakes with Simplified
Control Technique. In: SAE International Congress and Exposition; 28
February–4 March. Detroit, MI: SAE International; 1983. doi:10.4271/
830484.
[13] Vollrath M, Schleicher S, and Gelau C. The Influence of Cruise Control and
Adaptive Cruise Control on Driving Behaviour—A Driving Simulator Study.
Accident Analysis & Prevention. 2011; 43(3): 1134–1139. doi:10.1016/j.aap.
2010.12.023.
[14] Hall-Geisler K. How Electronic Stability Control Works 2021 [cited 11 May
2021]. Available from: https://auto.howstuffworks.com/car-driving-safety/
safety-regulatory-devices/electronic-stability-control.htm.
[15] Kortli Y, Marzougui M, and Atri M. Efficient Implementation of a Real-Time
Lane Departure Warning System. In: 2016 International Image Processing,
Applications and Systems (IPAS); 2016. p. 1–6. doi:10.1109/IPAS.2016.788
0072.
[16] Wang C, Sun Q, Li Z, et al. A Forward Collision Warning System Based on
Self-Learning Algorithm of Driver Characteristics. Journal of Intelligent &
Fuzzy Systems. 2020; 38(2): 1519–1530. doi:10.3233/JIFS-179515.
[17] Luo H, Yang Y, Tong B, et al. Traffic Sign Recognition Using a Multi-Task
Convolutional Neural Network. IEEE Transactions on Intelligent
Transportation Systems. 2018; 19(4): 1100–1111. doi:10.1109/TITS.2017.
2714691.
[18] Martinelli NS and Seoane R. Automotive Night Vision System. In:
Thermosense XXI. vol. 3700. International Society for Optics and Photonics.
SPIE; 1999. p. 343–346. doi:10.1117/12.342304.
[19] Centers for Disease Control and Prevention (CDC). Increasing Alcohol
Ignition Interlock Use; 2016 [cited 28 April 2021]. Available from: https://
www.cdc.gov/transportationsafety/impaired_driving/ignition_interlock_states.
html.
[20] Fuller J. How Pre-Collision Systems Work; 2009 [cited 28 April 2021].
Available from: https://auto.howstuffworks.com/car-driving-safety/safety-
regulatory-devices/pre-collision-systems.htm.
[21] Jacobé de Naurois C, Bourdin C, Stratulat A, et al. Detection and Prediction
of Driver Drowsiness Using Artificial Neural Network Models. Accident
Analysis & Prevention. 2019; 126: 95–104. 10th International Conference on
Managing Fatigue: Managing Fatigue to Improve Safety, Wellness, and
Effectiveness. doi:10.1016/j.aap.2017.11.038.
32 The Internet of medical things

[22] Atzori L, Iera A, and Morabito G. The Internet of Things: A survey. Computer
Networks. 2010; 54(15): 2787–2805. doi:10.1016/j.comnet.2010.05.010.
[23] Davoli L, Belli L, Cilfone A, et al. Integration of Wi-Fi Mobile Nodes in a
Web of Things Testbed. ICT Express. 2016; 2(3): 96–99. Special Issue on ICT
Convergence in the Internet of Things (IoT). doi:10.1016/j.icte.2016.07.001.
[24] Atzori L, Iera A, and Morabito G. From “Smart Objects” to “Social Objects”:
The Next Evolutionary Step of the Internet of Things. IEEE Communications
Magazine. 2014; 52(1): 97–105. doi:10.1109/MCOM.2014.6710070.
[25] Belli L, Cilfone A, Davoli L, et al. IoT-Enabled Smart Sustainable Cities:
Challenges and Approaches. Smart Cities. 2020; 3: 1039–1071. doi:10.3390/
smartcities3030052.
[26] Belli L, Cirani S, Davoli L, et al. An Open-Source Cloud Architecture for Big
Stream IoT Applications. In: Podnar Žarko I, Pripužić K, and Serrano M,
editors. Interoperability and Open-Source Solutions for the Internet of Things:
International Workshop, FP7 OpenIoT Project, Held in Conjunction with
SoftCOM 2014, Split, Croatia, September 18, 2014, Invited Papers. Springer
International Publishing; 2015. p. 73–88. doi:10.1007/978-3-319-16546-2_7.
[27] Welch KC, Harnett C, and Lee YC. A Review on Measuring Affect with
Practical Sensors to Monitor Driver Behavior. Safety. 2019; 5(4): 72.
doi:10.3390/safety5040072.
[28] Li M and Narayanan S. Robust ECG Biometrics by Fusing Temporal and
Cepstral Information. In: 2010 20th International Conference on Pattern
Recognition; 2010. p. 1326–1329. doi:10.1109/ICPR.2010.330.
[29] Abo-Zahhad M, Ahmed SM, and Abbas SN. Biometric Authentication
Based on PCG and ECG Signals: Present Status and Future Directions.
Signal, Image and Video Processing. 2014; 8(4): 739–751. doi:10.1007/
s11760-013-0593-4.
[30] Cassani R, Falk TH, Horai A, et al. Evaluating the Measurement of Driver
Heart and Breathing Rates from a Sensor-Equipped Steering Wheel Using
Spectrotemporal Signal Processing. In: 2019 IEEE Intelligent Transportation
Systems Conference (ITSC); 2019. p. 2843–2847. doi:10.1109/ITSC.2019.
8916959.
[31] Lanatà A, Valenza G, Greco A, et al. How the Autonomic Nervous System
and Driving Style Change With Incremental Stressing Conditions During
Simulated Driving. IEEE Transactions on Intelligent Transportation
Systems. 2015; 16(3): 1505–1517. doi:10.1109/TITS.2014.2365681.
[32] Eilebrecht B, Wolter S, Lem J, et al. The Relevance of HRV Parameters for
Driver Workload Detection in Real World Driving. In: 2012 Computing in
Cardiology. Krakow, Poland; 2012. p. 409–412.
[33] Samarin N and Sannella D. A Key to Your Heart: Biometric Authentication
Based on ECG Signals. CoRR. 2019; abs/1906.09181. Available from:
http://arxiv.org/abs/1906.09181.
[34] Coughlin JF, Reimer B, and Mehler B. Monitoring, Managing, and
Motivating Driver Safety and Well-Being. IEEE Pervasive Computing.
2011; 10(3): 14–21. doi:10.1109/MPRV.2011.54.
 Noninvasive psychophysiological driver monitoring 33

[35] Konieczka A, Michalowicz E, and Piniarski K. Infrared Thermal Camera-

based System for Tram Drivers Warning About Hazardous Situations. In:
2018 Signal Processing: Algorithms, Architectures, Arrangements, and
Applications (SPA); 2018. p. 250–254. doi:10.23919/SPA.2018.8563417.
[36] Cardone D, Perpetuini D, Filippini C, et al. Driver Stress State Evaluation by
Means of Thermal Imaging: A Supervised Machine Learning Approach
Based on ECG Signal. Applied Sciences. 2020; 10(16): 5673. doi:10.3390/
app10165673.
[37] Kiashari SEH, Nahvi A, Bakhoda H, et al. Evaluation of Driver Drowsiness
Using Respiration Analysis by Thermal Imaging on a Driving Simulator.
Multimedia Tools and Applications. 2020; 79(25–26): 17793–17815.
doi:10.1007/s11042-020-08696-x.
[38] Kashevnik A, Kruglov M, Lashkov I, et al. Human Psychophysiological
Activity Estimation Based on Smartphone Camera and Wearable
Electronics. Future Internet. 2020; 12(7). doi:10.3390/fi12070111.
[39] Lashkov I and Kashevnik A. Smartphone-Based Intelligent Driver Assistant:
Context Model and Dangerous State Recognition Scheme. In: Bi Y, Bhatia
R, and Kapoor S, editors. Intelligent Systems and Applications. vol. 1038.
Cham: Springer International Publishing; 2020. p. 152–165. doi:10.1007/
978-3-030-29513-4_11.
[40] Lindow F and Kashevnik A. Driver Behavior Monitoring Based on
Smartphone Sensor Data and Machine Learning Methods. In: 2019 25th
Conference of Open Innovations Association (FRUCT). Helsinki, Finland;
2019. p. 196–203. doi:10.23919/FRUCT48121.2019.8981511.
[41] Equivital Ltd. Eq LifeMonitor; 2021 [cited 18 May 2021]. Available from:
https://www.equivital.com/products/eq02-lifemonitor.
[42] Teledyne FLIR LLC. FLIR One Pro LT; 2021 [cited 18 May 2021].
Available from: https://www.flir.it/products/flir-one-pro-lt/.
[43] Laborde S, Mosley E, and Thayer JF. Heart Rate Variability and Cardiac
Vagal Tone in Psychophysiological Research – Recommendations for
Experiment Planning, Data Analysis, and Data Reporting. Frontiers in
Psychology. 2017; 8: 213. doi:10.3389/fpsyg.2017.00213.
[44] Malliani A, Pagani M, Lombardi F, et al. Cardiovascular Neural Regulation
Explored in the Frequency Domain. Circulation. 1991; 84(2): 482–492.
doi:10.1161/01.CIR.84.2.482.
[45] Shaffer F, McCraty R, and Zerr CL. A Healthy Heart Is Not a Metronome:
An Integrative Review of the Heart’s Anatomy and Heart Rate Variability.
Frontiers in Psychology. 2014; 5: 1040. doi:10.3389/fpsyg.2014.01040.
[46] ADInstruments Ltd. LabChart Lightning; 2021 [cited 10 May 2021].
Available from: https://www.adinstruments.com/products/labchart.
 Chapter 3
IoT-based biomedical healthcare approach
Anil Audumbar Pise1

The Internet of Things (IoT) enables physical items and devices to see, hear, and
think by exchanging data. The IoT works in the other direction, converting ordinary
items into intelligent ones. Embedded devices, communication protocols, sensor
networks, Internet protocols, and applications are examples of IoT-specific tech-
nology. Certain IoT-based healthcare solutions, such as mobile health and telecare,
as well as preventative, diagnostic, therapeutic, and monitoring systems, funda-
mentally modify everything. Wireless body area networks (WBANs) and radio-
frequency identification (RFID) are unquestionably important components of the
IoT. This chapter is covered, in addition to research on the usage of IoT in the field
of biomedical systems to be applied within the framework.

3.1 Introduction
The fast progress of IoT technology in recent years has permitted the connecting of a
huge number of smart items and sensors, as well as the formation of seamless data
interchange between them. As a result, data analysis and storage systems such as
cloud computing and fog computing are required. Healthcare is one of the IoT
application areas that has aroused the interest of industry, academia, and government.
The advancement of IoT and biomedical healthcare is enhancing patient
safety, staff happiness, and operational efficiency in the medical industry.
Biomedical engineering is a relatively new area of study that integrates funda-
mental principles from physical and applied science to life and medicine. Engineers
and experts in biomedical engineering work at the interface of engineering,
healthcare and biological sciences. Recent advancements in the semiconductor
industry, such as the down-scaling of electrical devices, as well as advancements in
computer sciences, such as the advent of data science and cloud computing, have
sparked an explosion of novel therapeutic applications. The effect was slow, but it
helped the whole field of medicine. Nowadays, healthcare tailored to individual
pathological requirements, the retention of long-term data (ECG, SPO2, etc.), and

1
School of Computer Science and Applied Mathematics, University of the Witwatersrand, Johannesburg,
South Africa
36 The Internet of medical things

Internet

Long-term data Biosignal processing

Physiological signal

Heart
ECG
ABP Feedback
Arteries
PPG Classification and analysis

Figure 3.1 IoT-based healthcare equipment in biomedical approach [1]

analysis to identify pathological indications (blood pressure, perfusion index, glu-

cose level, etc.). In general, monitoring patients with chronic illnesses continuously
and in real time has become simpler than ever and we will soon be able to foresee
illness.
The previous diagram (Figure 3.1) depicts a simplified form of an IOT-based
biological system that includes monitoring, storage, complicated decision-making,
and feedback capabilities. In most cases, a device is connected to the bodies of
participants in order to collect and transmit data. This device is made up of an analog
front-end circuit with sensors, amplifiers and filters for capturing physiological sig-
nals (such as ECG, PPG, and EEG), a processing unit, and a communication module.
As a consequence of improvements in microelectronics technology, we now have
wearable and implanted biosensors. The majority of long-term data is delivered
through the Internet to cloud storage to alleviate the strain of processing huge
amounts of data acquired from individuals and to conserve energy. The advent of
“IoT” technology has made such connections between the front-end module and
storage simpler than before.
A server with biological signal-processing capabilities may therefore execute
primary processing tasks such as noise reduction (filtering) and sickness identifica-
tion (QRS complex detection, R–R interval measurement, etc.). Furthermore, a
“machine learning” and “big data” method may be used to make more sophisticated
decisions (e.g., aberrant heart activity categorization), identify illnesses (e.g., cardiac
disease identification), and detect dangerous situations early (e.g., stroke prediction).
Finally, the results of this analysis may be made accessible to the general public,
authorized physicians, and clinics through the mobile application, in certain cir-
cumstances nearly immediately after the device is connected to the subject’s body.
In the remainder of this chapter, the following order has been created. In
Section 3.2, a brief explanation of IoT-based healthcare biomedical application is
 IoT-based biomedical healthcare approach 37

given, which is followed by an overview of IoT-based healthcare biomedical sys-

tem. Section 3.3 explores the IoT-based biomedical communication architecture
and the healthcare system’s interconnections. In Section 3.4, WBANs in IoT
healthcare are explained. In Section 3.5 RFID communication protocol for bio-
medical applications is explained. A brief synopsis of analysis related to this study
is explained in Section 3.6. This section discusses how IoT is being used to improve
the delivery of biomedical operations and healthcare services during a time of
crisis. The conclusion and potential work in the following sections are provided.

3.2 IoT-based healthcare biomedical applications

Biomedical applications powered by the IoT are used in biomedical systems for
healthcare/telecare, diagnosis, prevention, treatment, and monitoring. WBANs and
RFID technologies are key to the IoT concept. The goal of this chapter is to explore
a new IoT-based healthcare framework for hospital information systems that will
make use of WBANs and RFID technologies. The IoT-based biomedical frame-
work is modeled and simulated using Riverbed Modeler. The results reveal that the
IoT-based biomedical framework energy-aware system fulfills the ISO/IEEE
11073 standard’s data rate and latency QoS criteria. It is also shown that the pro-
posed framework may be utilized to swiftly develop unique case studies in
healthcare information systems by building a time-saving simulation environment.
This study has looked at the most recent IoT components, applications and
industry breakthroughs in healthcare, as well as the current status of IoT and bio-
medical healthcare apps, since 2015. In addition, to look at how promising tech-
nologies like biomedical healthcare systems, ambient supported living, big data,
and wearables are being applied in the healthcare business. Furthermore, the dif-
ferent IoT and e-health legislation and policies in existence throughout the globe
will be examined to evaluate their influence on the long-term growth of IoT and the
healthcare business. Furthermore, a thorough assessment of IoT privacy and
security issues from a healthcare standpoint is performed, including potential
threats, attack techniques, and security setups. Finally, this chapter covers pre-
viously well-known security solutions in order to solve security issues, as well as
trends, highlighted potential, and issues for the future evolution of IoT-based bio-
medical healthcare systems.
Though research in adjacent fields has shown that remote health monitoring is
feasible, the potential benefits in a variety of circumstances are significantly
greater. Remote health monitoring might be used to follow noncritical patients at
home rather than in the hospital, alleviating pressure on hospital services such as
physicians and beds. It might be used to improve access to healthcare in remote
locations or to assist elderly people in remaining independent at home for long
periods of time. Essentially, it may improve access to professional services, alle-
viating load on healthcare facilities and empower individuals to have more control
over their own health at all times. Figure 3.2 depicts a typical IoT-based biomedical
healthcare system.
38 The Internet of medical things

Outpatient care
EHR systems

Electronic
medical consultation

Smart emergency

Network-based
community- EHR
Internet of
Things

Healthcare PAN

Elderly monitoring
Smart medicine
Heart Rate
sensor

Motion
sensor

Figure 3.2 Overview of a typical IoT-based biomedical healthcare system [2]

3.3 IoT-based biomedical communication architecture

In research [3], Chiang et al. developed a robust nursing-care support system to

enable efficient and secure communication between mobile biosensors, active
intelligent objects, the IoT gateway, and the backend nursing-care server, which
can do further data analysis to deliver high-quality, on-demand nursing care. This is
an IoT-based biomedical communication framework to improve healthcare
services.
In Figure 3.3, a patient management and rehabilitation scenario is shown in
which field-installed environmental sensors and medical sensing devices serve as
the IoT connection architecture for a nursing-care support system [3]. This enables
caregivers (such as nurses) to perform patient care and rehabilitation procedures
more effectively. Three major components comprise the proposed IoT connection
scenario: a backend nursing-care server, a mobile gateway (usually carried by the
caregiver), and intelligent gadgets (such as fixed sensor nodes or medical sensing
devices). The intelligent gadgets will monitor and collect environmental and patient
 IoT-based biomedical healthcare approach 39

Patient management

Nursing-care
server
Nurses with
mobile gateway

Rehabilitation

Smart biosensor
Fixed environmental sensor

Figure 3.3 IoT-based biomedical communication architecture [3]

biodata, with the caregiver engaging with them through a portable smart smartphone.
The patient gives biometric data, namely, electrocardiography, electroencephalography,
electromyography, and blood pressure. Following that, the caregiver may combine
ambient data and the patient’s biodata to make a more rapid assessment of the patient’s
requirements. As a consequence, patients will get more precise and timely therapeutic
intervention. Numerous studies have been conducted on physical injury healing for
enhancement of biomedical applications using IoT. In [4], Fan et al. describe the process
of creating a framework that culminates in a recovery strategy personalized to an indi-
vidual’s symptoms. This is accomplished by linking the patient’s diagnosis to a history of
past patients’ symptoms, illnesses, and therapies. The technology allows doctors to
manually input symptoms and approve prescription medication; in 87.9% of cases. The
physician followed the system’s instructions to the letter and made no changes to the
recommended treatment plan.

3.4 Wireless body area networks

The WBAN is made up of implanted sensors that monitor vital indicators as well as
specific organs inside the body. Sensors capture data from various parts of the body
and transmit it to the sink. Recent research has concentrated on the dependability,
security, and efficiency of sensor communications in order to convey data from
sensors to sinks. Reduced network life span is a critical criterion in WBANs, and
the majority of recent publications have focused on developing cooperative routing
algorithms that achieve low-energy consumption and increased network longevity.
The packet reception ratio, on the other hand, is another statistic that has been used
into the routing algorithms of WBANs in order to determine whether or not packets
were successfully received in the sink [5].
40 The Internet of medical things

Reference [6] assesses the usefulness of existing IoT technologies as a tool for
monitoring Parkinson’s disease patients. Their research implies that wearable
monitors capable of detecting gait patterns, tremors, and general activity thresholds
might be used in conjunction with vision-based sensors (i.e., webcams) in the home
to detect Parkinson’s disease beginning. Additionally, the authors believe that
machine learning might result in more successful treatment options in the long run.
According to [7], WBAN is on the way to improve biomedical application based on
IoT. It consists of a network of small intelligent devices that can communicate
wirelessly and analyze data. The garment incorporates sensing knots and may be
worn as clothes. There are numerous things that are unique or may be worn over
others. The growth of technology has resulted in a reduction in the size of elec-
tronic devices capable of wireless communication, while increasing their proces-
sing and data storage capabilities. WBANs have become a crucial component of
medical monitoring systems due to their ability to be used at any time and in
any place.
Figure 3.4 displays the fundamental WBAN structure as well as information
transmission strategies collected from individuals. This structure’s first layer
facilitates wireless communication, allowing sensor nodes and nodes to connect
with the network structure on the higher layer [9]. A Wi-Fi gateway is provided
(PDA, smart mobile phone, etc.) for communication. The gateway is often a
computer with an Internet connection that is attached to an access point and wire-
lessly captures data sent over it. The data gathered is kept on this computer.
Currently, a medical monitoring mechanism is in place, enabling all data to be
analyzed. It is preserved in such a manner that authorities may access it remotely.

EEG
Vision sensor

Temperature sensor

ECG
Breathing sensor

Blood pressure
Glucose level
Network

Motion sensor

Figure 3.4 General architecture of WBAN [8]

 IoT-based biomedical healthcare approach 41

Since the primary goal of WBAN is to improve people’s lives, several appli-
cations are accessible. For the most part, these applications may be classified as
either medical or nonmedical. Monitoring physiological aspects of the human body
in order to identify anomalies and warn relevant people in real-time is one example
of a medical application. Nonmedical uses include entertainment, emotion detec-
tion, secure identification, and nonmedical emergencies that are accomplished by
collecting environmental data and alerting people to hazards like as fire [10]. While
WBAN systems are often used to monitor a single person, collaborative WBAN
systems may be used to monitor a group of people.

3.5 RFID IoT-based biomedical communication protocol

RFID technology offers a broad variety of uses in daily life, particularly where
security is of the utmost importance. Monitoring human behaviors is one
example of a mission-critical application that prioritizes security in biomedical
healthcare applications. Due to population growth and random mobility, it is
conceivable to encounter an inability to distinguish between two or more RFID-
tagged individuals in a guarded environment with an RFID-based people mon-
itoring system in place. One of the key challenges that security professionals
confront in this regard is quickly recognizing and counting the number of tags
present in a monitoring area.
In this context, a simple method was thought for avoiding RFID tag collisions
in a busy, chaotic setting. This technique is known as the “adaptive slot adaptive
frame (ASAF)-ALOHA protocol.” This chapter briefly explores the ASAF-
ALOHA protocol as an alternative to the presently utilized MAC protocols to
enhance the overall performance of the RFID system. The ASAF-ALOHA protocol
does a probabilistic assessment of the number of active tags in the region of interest
and modifies the frame size and slot count in the next round accordingly. Because it
conveniently handles RFID multiple-tag anti-collision problems, the adaptive
slotted ALOHA protocol was selected as the communication layer for the case
explored in this study. It is also worth noting that the physical layer communicates
at a frequency of 13.56 MHz.
This is accomplished by the use of a simple algorithm that assesses the sys-
tem’s performance using statistical data on the number of successful, failed, and
idle slots available to the reader. ASAF-ALOHA has been demonstrated in
experiments to outperform conventional MAC procedures. In actuality, many
algorithms’ versatility is constrained by slots or frames. The ASAF-ALOHA pro-
tocol, on the other hand, is more flexible and allocates time for tag detection by
distributing many frames in each round with time-adjustable slots per frame; that is,
both the round and frame sizes may be changed at the same time. This improves the
chances that the object will be discovered. The application layer, communication
layer, and physical layer are the three levels of a simple RFID communication
protocol. Figure 3.5 depicts a typical RFID-based healthcare system using IoT-
based biomedical application.
42 The Internet of medical things

Users RFID-enabled Tagged objects

healthcare systems

Asset tracking
system Drug
Patient container
Doctor Patient
identification
system RFID
tag
Drug
administration Medical
Asset supplies
system
Nurse

Access control
system

RFID readers
Patient Terminal

Internet/
wireless network

Database Hospital legacy

server system

Figure 3.5 RFID-based healthcare system for IoT analytics in medicine

The author of [11] suggested a simple method for diabetics to assess their
blood glucose levels. Patients must manually measure blood glucose levels at
certain intervals during this technique. It next goes through two types of blood
sugar irregularities. The first includes increased blood glucose levels, while the
second includes a failure to assess blood glucose levels. Furthermore, the system
determines how to tell the patient, his or her parents and family friends, or emer-
gency healthcare professionals such as physicians, dependent on the severity of the
irregularity. While this methodology is feasible and has been shown to be feasible,
it might be improved by automating blood glucose monitoring.
In authors’ suggestions [12], an off-the-shelf component and a custom antenna
were combined to create a device for predicting cardiac disorders. A micro-
controller analyses data from an ECG sensor to identify the heart rhythm.
This data is then wirelessly transferred to the user’s device through Bluetooth,
where it is processed and presented in a user application. The authors suggest that
creating algorithms to forecast heart attacks might help the approach. Additional
progress might be achieved by incorporating respiration rate monitoring, which has
been shown to aid in the detection of cardiac arrests [13].
 IoT-based biomedical healthcare approach 43

3.6 Problem statement

Much of the current research on IoT in healthcare has focused on the many uses of
biomedical technology in a number of healthcare settings, including as nursing,
ambient assisted living, and surgery. Furthermore, no studies have been undertaken
that focused only on IoT improvements in healthcare and then compared the results
to other healthcare businesses. This chapter, as an extension of prior study, presents
an in-depth assessment of the numerous categories of specific elements of IoT
advancements in medicine, with the purpose of giving this information to stake-
holders interested in these types of innovations.

3.7 IoT and biomedical healthcare system

interconnection
Previously, there was just one way for patients to communicate with doctors: in
person, over the phone, or through text messaging. There was no system in place for
doctors and hospitals to periodically evaluate patient health and, as a result, offer
appropriate medications. The ability of the IoT to link technology and people’s
physical bodies will almost surely transform the healthcare business. Patients, family
members, professionals, hospitals, and the healthcare system as a whole gain from it.
Innovative technologies that aid patients and doctors in ensuring their safety
and enhancing treatment via the use of smart devices enabled by the IoT-based
biomedical instruments offer up new paths for medical monitoring. At the same
time, more open physician–patient communication has increased patients’ partici-
pation and satisfaction. Furthermore, patients’ health may be monitored, minimiz-
ing the duration of their stay and the possibility that they would return to the
hospital following release. The widespread use of the IoT has the potential to
reduce healthcare costs while boosting treatment efficacy. Furthermore, [14]
includes statistical models for calculating joint angles in physical hydrotherapy
scenarios, allowing for the monitoring of joint activity improvement over time.
Simple mechanical and electrical components have evolved into complicated
devices that involve software, with some growing into platforms [15]. It may be
observed in healthcare systems that have gotten smarter as a result of the IoT
technology. Chronic illnesses may be tracked remotely and measures exchanged
instantaneously with physicians and patient families thanks to the IoT-based bio-
medical architecture. This enables for regular monitoring of measurements as well
as the early discovery and avoidance of unpleasant disease-related circumstances.

3.8 Examples of IoT-based biomedical

healthcare devices
Initiatives to use the IoT-based biomedical healthcare devices in medicine are
being developed. The effect of restricted healthcare resources on the lab technician
44 The Internet of medical things

workforce, as well as the decrease in healthcare staff due to the deployment of

health-related IoT-based biomedical devices, promises to reduce some of the stress
associated with the blood shortage. Invasive, transdermal medications, pacer sys-
tems, electronic gauges, and other forms of drug monitors allow doctors to help
patients with infections and check on their well-being while also aiding with
therapy. Growth, on the other hand, offers a myriad of advantages as well as a
multitude of downsides. Many people have raised these self-evident healthcare
security issues. Data thieves have access to data gathered and stored by embedded
devices in the IoT. Prior to broad use, additional expansion of the usage of Internet
of Healthcare devices and networks is necessary. Cybersecurity is a major barrier to
a broad use of the IoT in healthcare.

3.8.1 Glucose monitoring

Glucose control has long been a struggle for the more than 30 million diabetics in
the United States. Manually monitoring and documenting glucose levels is not only
time-consuming, but it also only captures a patient’s glucose levels at the time of
the test. When levels fluctuate drastically, standard monitoring may be inadequate
to identify a problem.
IoT technologies, which provide continuous, automated monitoring of patients’
glucose levels, may assist to alleviate these worries. Glucose monitoring systems do
away with the necessity for manual record-keeping and alert patients when their
glucose levels are abnormal. Figure 3.4 depicts wireless implantable medical devices
based on the IoT.
Among the issues is creating a glucose monitoring system based on the IoT as
follows:
● Small enough to be used surreptitiously over time to observe patients without
causing pain.
● It does not use a lot of energy and hence has to be recharged on a regular basis.
These are not insurmountable obstacles, and technologies that solve them have
the potential to change the way individuals regulate their glucose levels.

3.8.2 Bluetooth-enabled blood labs

Implantable laboratories at the Swiss Federal Institute of Technology in Lausanne
enable an independent analysis of patients’ own blood samples, giving critical
information. The unit consists of five electrodes, each coated with an enzyme, which
enables the implant to monitor substances such as glucose and lactate. After finding
the material, it is scanned and wirelessly sent through Bluetooth to the individual’s
PC. Additional investigation may be necessary; however, the results of this study
may be sent through a smartphone to a doctor in another location for examination.
This development would mostly benefit the elderly and chronically sick, since it
would decrease their need on physicians. When implanted, this implantable device
reduces the need for further blood work, allowing lab workers to be reduced. A
completely automated laboratory that removes face-to-face interaction with patients
 IoT-based biomedical healthcare approach 45

results in decreased time spent acquiring all necessary lab tests. Additionally, they
are advantageous for a number of reasons, including relieving demand on an over-
burdened healthcare system.

3.8.3 Connected inhalers

Asthma is a huge condition that affects hundreds of millions of individuals
throughout the world. Asthmatics may boost their reservoirs by using connected
inhalers, giving them greater control over their symptoms and therapy. They
employ cutting-edge asthma software. Propeller has developed a sensor that can
wirelessly interact with an inhaler or spirometer. The sensor is designed to alert
patients with asthma and chronic obstructive pulmonary disease to their thoughts as
well as information that may help them make decisions about their worries. The
sensor and related software detect medicine consumption and allergy threats as well
as predict and notify users of these changes. Some saw the linked inhaler as a big
advantage, despite the fact that it requires more effort on the patient’s part to use.
Furthermore, the sensor generates a report that may be shared with the patient’s
physician.

3.8.4 Blood coagulation testing

Roche has launched a Bluetooth-enabled coagulation instrument. Using the IoT
system, patients may track the rate at which their blood clots. Roche’s technology is
the first IoT platform created specifically for anticoagulated patients. Self-testing
has been shown to help patients remain within their treatment range and reduce
their risk of bleeding and stroke. Because of the capacity to wirelessly transfer test
results to their healthcare practitioner, patient visits have dropped. Furthermore,
Roche’s device enables patients to comment their test results, reminds them to test,
and alerts them when test results exceed a specified range.

3.8.5 Connected cancer treatment

The American Society of Clinical Oncology’s (ASCO) 2018 Annual Meeting will
include a presentation of the findings of a clinical research involving patients
receiving treatment for head and neck cancer. Patients will wear Bluetooth
devices while receiving therapy, like in the pilot experiment, and their pain, heart
rate, and even weight will be gathered through Bluetooth and continually mon-
itored using data collection software coupled with the blood pressure and pulse-
tracking program. In addition, each patient’s weight will be measured on a daily
basis to provide crucial information to their doctor. If the doctor believes it is
necessary, the medicine may be changed on a daily basis. When compared to
individuals in the placebo group who got no medical services connected to cancer
and those who got medical services once a week, study participants saw a
reduction in their symptoms.
The use of technology, which enabled the recognition of adverse effects
associated with advanced healthcare practices and the implementation of smart
health practices, has aided in simplifying patients’ treatment by assisting with any
46 The Internet of medical things

emerging side effects, as well as addressing and alleviating concerns about it.
Better treatments with smart technology may reduce patients’ inconveniencing.

3.8.6 Robotic surgery

Surgeons may perform intricate procedures that would be impossible with human
hands by implanting miniature Internet-connected robots inside the human body.
Simultaneously, robotic surgeries done using small IoT devices would substantially
reduce the number of surgical incisions, resulting in less painful therapy and a
faster recovery period for patients.

3.8.7 IoT-connected contact lenses

Lens expansion and lens-stiffening treatments such as extracapsular categorization
are thought to be the definitive solution for acquired longness, a cataract ophthal
eye ailment. A study targeted at developing technology to treat long-lens failure,
also known as presbyopia, the condition that leads in long-flectomizedness, will
look at lens healing/stiffening and refracting to ensure recovery. Sensimed signals,
developed by Swiss researchers, detect variations in ocular pressures that may
indicate glaucoma.

3.8.8 A smartwatch app that monitors depression

Patients with MDD may need to wear smartwatches and utilize a monitoring gadget
to track their depressive symptoms on a frequent basis. In this scenario, wearable
technology has a far better chance of making a difference than just collecting steps;
sensors that detect depression severity would be ideal. A depression app, like other
IoT apps, may provide patients and caregivers with more information about their
challenges.

3.8.9 Connected wearables

The IoT seems to be incomplete without interconnected wearables. Because worn and
connected sensors are valuable equipment for health and medical staff, they not only
benefit but also aid patients. They enable medical staff to monitor critical body sta-
tistics such as heart rate, body temperature, pulse, and other vital body data while still
analyzing patients’ health. Furthermore, hospitals benefit from wearable technology
that allows doctors to monitor patients even after they leave the clinic. It is particularly
useful in assisting patients with frequent checkup visits if they address problems they
encountered during their hospital stay following release. If the situation changes,
wearable sensors will alert the doctor from anywhere and at any time. When doctors
are notified in real time, they may provide real-time guidance to their patients.

3.9 A summary of associated research

The IoT is rapidly gaining significance owing to its significant benefits of greater
precision, reduced costs, and the ability to forecast future occurrences. Despite the
 IoT-based biomedical healthcare approach 47

fact that smartphone and computer technology have been in use for a long time, the
increasing availability of apps and the IoT-based biomedical applications, wireless
technology, and the virtual economy has all contributed to the rapid IoT, causing
the overall technology ecosystem of society to expand [16].
Other physical instruments (sensors, actuators, and so on) have been com-
bined with IoT devices to gather and communicate data through Bluetooth, Wi-
Fi, and IEEE 802.11 [17]. Heart-related applications incorporate clinical data
about the patient using integrated or wearable sensors such as thermistors, a
palpator (a pressure sensor), a lead electrograph physiologist’s (an ECG), and
electro-of pathologist’s the brain’s electric potential (an EEG). Environmental
parameters that may be recorded include temperature, humidity, date, period,
and time of day. These medical information may be used to make intriguing and
precise conclusions about a patient’s clinical status. Because IoT sensors gen-
erate/capture a variety of data through the Internet, enormous amounts of data
are provided by a variety of sources (sensors, mobile phones, e-mail, software,
and applications). The previous research’s findings are made accessible to phy-
sicians, caregivers, and anyone else allowed to use the devices. The growth of
cloud/server-based stile diagnosis and the transmission of these facts through
healthcare services is an efficient use of information, and cures are delivered as
soon as possible when required.
Because the app’s users, the healthcare institution, and the contact module all
function together, all parties have access to the app’s data. The bulk of IoT com-
ponents are user-interface-based (IoT in the framework largely functions as a
dashboard for medical practitioners, enabling patient monitoring, data visualiza-
tion, and apprehension capabilities). The research revealed that IoT in healthcare
has improved on previous year’s results by delving further into these issues [14].
Healthcare monitoring, regulation, and privacy are three potential IoT applications
that should be examined first. These technical advances demonstrate the IoT
potential for success and profit in the healthcare business. The fundamental pro-
blem, however, is to maintain service quality matrices that encourage knowledge
exchange, stability, cost efficiency, and resilience while also maintaining all users’
data privacy [18].
Table 3.1 summarizes the contributions of several academics to IoT in
healthcare. In 2019, Dang et al. developed a cloud-based method for processing IoT
data in healthcare [24]. They selected to address a number of security problems and
difficulties related to the use of IoT and cloud computing in healthcare. In 2019
[27] and 2018 [23,28], a significant number of peer-reviewed journal articles were
released, both of which searched for prior-to-2017 publications. Nazir et al. per-
formed a systematic review of research published between 2011 and 2019 that
addressed the security and privacy risks associated with mobile-based healthcare
IoT [27].
There are presently no research papers that are solely devoted to medical IoT
applications; nonetheless, Table 3.1 provides relevant research papers on healthcare
IoT. The table summarizes the outcomes of ten independent researches on IoT in
healthcare (e.g., systemic analyses and other types of evaluations). The initial
48 The Internet of medical things

Table 3.1 General review of IoT in healthcare

Paper title Contributions Author and

reference
The IoT for health care: a compre- ● Discuss the problems and con- Islam et al. [19]
hensive survey cerns associated with the Internet
of Things in healthcare
● Address different laws and legis-
lation governing IoT and
eHealth. In healthcare, discuss
big data, ambient knowledge, and
wearables
● Examine network structures and
associated software or facilities
in IoT for healthcare
● Emphasize the protection and
privacy concerns associated with
IoT in the healthcare network
Medical internet of things and big ● Examine the use of IoT and big Dimitrov [20]
data in healthcare data in healthcare
● Discuss the problems associated
with the use of big data in
healthcare
● Apps and smartphone applica-
tions for examination
Internet of things for smart health- ● A definition for the usage of IoT Baker et al. [21]
care: technologies, challenges, in healthcare is proposed
and opportunities ● The current state of affairs and
possible future advances in
healthcare IoT are addressed
● Discuss potential research trends,
concerns, and issues of health-
care’s Internet of Things
● Consider cloud computing as a
data storage system
● Pay attention to different wear-
able and networking gadgets
Advanced internet of things for ● Examine and categorize health- Yang et al. [22]
personalised healthcare system: a care IoT systems, implementa-
survey tions, and successful case studies
● Present a four-layer Internet of
Things architecture for custo-
mized healthcare networks (PHS)
● Discuss upcoming study devel-
opments, as well as problems and
obstacles in healthcare IoT
(Continues)
 IoT-based biomedical healthcare approach 49

Table 3.1 (Continued)

Paper title Contributions Author and

reference
● Provide an overview of the pre-
sent situation and potential
developments of healthcare IoT
Towards fog-driven IoT eHealth: ● Describe a multi-tiered archi- Farahani et al.
promises and challenges of IoT in tecture for the Internet of Things- [23]
medicine and healthcare enabled eHealth ecosystem: To
inspire, use a Devices, fog com-
puting, and a cloud-based service
● Discuss how the Internet of
Things will be seen in hospitals
and pharmacy
● Showcase a unified platform for
the Internet of Things-enabled
eHealth world
● Demonstrate the challenges and
future avenues for IoT in e-health

research [19] examined several aspects of IoT in healthcare, including networks

and structures, methods, and implementations. They discussed a wide range of
issues, including data privacy and standards, as well as IoT and e-health regula-
tions. On the other hand, the contents of this study date back to the early days of
IoT, and the need for it in the healthcare industry is expanding, forcing the creation
of a fresh assessment.
Table 3.2 provides a thorough [28] overview of the materials and technology used
in hospital IoT installations reported in 2018. According to the report, the most com-
mon site for healthcare IoT is the home. It demonstrates that the top-ranked journals for
papers created on the subject are Procedia Computer Science, the Journal of Network,
and Computer Applications, and the Journal of Medical Systems. Another research [29]
from 2017 revealed that hospitals were the most vulnerable to IoT adoption. Certain
countries’ e-health programs and activities have been touted as contributing to the IoT
without providing readers with a summary of their work [19,24].
Another issue that IoT has presented is security and secrecy, as well as inter-
operability and interaction with healthcare technologies. In previous articles, we went
into further depth on these topics [19–21,23,25–28,30]. Previously, these academic
articles focused on various aspects of IoT healthcare, such as the use of cloud com-
puting, fog computing, mobile computing, wearable sensors, and big data in health-
care. Several research articles on the current state and potential breakthroughs for
healthcare IoT address current and future advances [21,23,25,27,28,30].
50 The Internet of medical things

Table 3.2 Systematic review of IoT in healthcare

Paper title Contributions Author and

reference
A survey on internet of things ● Conduct a study of the IoT framework Dang et al. [24]
and cloud computing for for the healthcare sector, with an
healthcare emphasis on architecture, platforms,
and topologies
● Examine the influence of IoT and cloud
computing on healthcare
● Describe emerging industry trends and
initiatives on IoT and cloud computing
in the healthcare field around the world
● Discuss the security issues associated
with the Internet of Things (IoT) and
cloud computing in healthcare
● Consider how the use of IoT and cloud
computing in healthcare poses a number
of challenges
The application of internet of ● Investigate the cybersecurity and inter- Ahmadi et al.
things in healthcare: a sys- operability issues involved with [28]
tematic literature review healthcare IoT
and classification ● Show the current state of healthcare IoT
and then present possible future
changes
● More about the Internet of Things and
how it could be included in healthcare
● Examine the advantages of cloud-based
architecture for IoT in healthcare
Internet of things for health- ● Show the current state of healthcare IoT Nazir et al. [27]
care using effects of mo- and then present possible future
bile computing: a changes
systematic literature re- ● Examine the security and privacy issues
view posed by healthcare IoT
● Mobile computing in healthcare IoT
IoT-based healthcare appli- ● Show the current state of healthcare IoT de Morais
cations: a review and then present possible future Barroca Filho
changes et al. [25]
● Examine the security and privacy issues
posed by healthcare IoT
● Mobile computing in healthcare IoT
Enabling technologies for fog ● The current state of affairs and possible Awad Mutlag
computing in healthcare future advances in healthcare IoT are et al. [26]
IoT systems addressed
● Collect both of their functioning and
non-functional requirements
● Discuss the issues and questions around
IoT infrastructure in healthcare
 IoT-based biomedical healthcare approach 51

3.10 Conclusion
One of the aims of e-health is to deliver healthcare services to patients in their
homes, especially via the use of IoT-based biomedical applications. In general, IoT
applications are intended to save money and stimulate patients at home, resulting in
increased patient engagement. Everyone will benefit from improved health pro-
motion and a more fulfilled lifestyle as a consequence of this.
According to the study’s findings, the IoT-based biomedical applications in
medicine are still in its infancy. The utilization-limiting application seems to have
put a significant strain on the healthcare system in a variety of subfields.
Furthermore, since the number of medical IoT studies and research fields increased
in 2018, the number of studies on this topic will grow in the future, resulting in the
engagement of new research areas.

3.11 Future work

The healthcare sector is a large and diverse enterprise in which a wide range of actors,
including patients, healthcare specialists, and insurance firms, are heavily involved.
Nonetheless, despite the fact that IoT is currently not extensively employed in a
variety of medical subfields, it is gaining traction in those sectors. Despite the fact that
there is much overlap among IoT, healthcare, and medicine, there are certain sectors
where IoT has yet to be applied due to interdisciplinary challenges. Scholars who are
particularly interested in comprehending this phenomenon may be interested in this
investigation. This goal will be met in the future either qualitatively, via interviews
with targeted audiences, or quantitatively, via literature searches.

References
[1] C.-F. Chiang, F.-M. Hsu, and K.-H. Yeh. Robust IOT-based nursing-care
support system with smart bio-objects. Biomedical Engineering Online, 17
(2): 1–15, 2018.
[2] Y.J. Fan, Y.H. Yin, L.D. Xu, Y. Zeng, and F. Wu. IoT-based smart rehabi-
litation system. IEEE Transactions on Industrial Informatics, 10(2): 1568–
1577, 2014.
[3] R. Jafari and M. EffatParvar. Cooperative Routing Protocols in Wireless
Body Area Networks (WBAN) : A Survey, 2017.
[4] C.F. Pasluosta, H. Gassner, J. Winkler, J. Klucken, and B.M. Eskofier. An
emerging era in the management of Parkinson’s disease: wearable technol-
ogies and the internet of things. IEEE Journal of Biomedical and Health
Informatics, 19(6): 1873–1881, 2015.
[5] I. Kırbas. Online Kablosuz İnkübatör İzleme ve Kontrol Sistemi Tasarımı ve
Uygu-laması. PhD thesis, Doktora Tezi, Sakarya Üniversitesi, Sakarya-
Türkiye, 2013.
52 The Internet of medical things

[6] S. Saleem, S. Ullah, and K.S. Kwak. A study of IEEE 802.15.4 security fra-
mework for wireless body area networks. Sensors, 11(2): 1383–1395, 2011.
[7] A. Pise, H. Vadapalli, and I. Sanders. Facial emotion recognition using
temporal relational network: an application to e-learning. Multimedia Tools
and Applications; Springer: Netherland, 2020; pp. 1–21.
[8] S.-H. Chang, R.-D. Chiang, S.-J. Wu, and W.-T. Chang. A context-aware,
interactive m-health system for diabetics. IT Professional, 18(3): 14–22,
2016.
[9] G. Wolgast, C. Ehrenborg, A. Israelsson, J. Helander, E. Johansson, and H.
Manefjord. Wireless body area network for heart attack detection [education
corner]. IEEE Antennas and Propagation Magazine, 58(5): 84–92, 2016.
[10] M.A. Cretikos, R. Bellomo, K. Hillman, J. Chen, S. Finfer, and A. Flabouris.
Respiratory rate: the neglected vital sign. Medical Journal of Australia, 188
(11): 657–659, 2008.
[11] R.C.A. Alves, L.B. Gabriel, B.T. de Oliveira, C.B. Margi, and F.C.L. dos
Santos. Assisting physical (hydro) therapy with wireless sensors networks.
IEEE Internet of Things Journal, 2(2): 113–120, 2015.
[12] A. Pise, H. Chabukswar, and D. Jadhav. Hard authentication using dot array
carrier structure. International Journal of Scientific & Engineering
Research, 4(5): 670–676, 2014.
[13] V. Jagadeeswari, V. Subramaniyaswamy, R. Logesh, and V. Vijayakumar. A
study on medical internet of things and big data in personalized healthcare
system. Health Information Science and Systems, 6(1): 1–20, 2018.
[14] H. Peng, Y. Tian, J. Kurths, L. Li, Y. Yang, and D. Wang. Secure and
energy-efficient data transmission system based on chaotic compressive
sensing in body-to-body networks. IEEE Transactions on Biomedical
Circuits and Systems, 11(3): 558–573, 2017.
[15] A. Gatouillat, Y. Badr, B. Massot, and E. Sejdić. Internet of medical things:
a review of recent contributions dealing with cyber-physical systems in
medicine. IEEE Internet of Things Journal, 5(5): 3810–3822, 2018.
[16] L.M. Dang, M. Piran, D. Han, et al. A survey on internet of things and cloud
computing for healthcare. Electronics, 8(7): 768, 2019.
[17] I. de Morais Barroca Filho and G.S. de Aquino Junior. IoT-based healthcare
applications: a review. In International Conference on Computational
Science and Its Applications, pages 47–62. Springer, 2017.
[18] A. Awad Mutlag, M.K. Abd Ghani, N. Arunkumar, et al. Enabling tech-
nologies for fog computing in healthcare IoT systems. Future Generation
Computer Systems, 90: 62–78, 2019.
[19] S. Nazir, Y. Ali, N. Ullah, and I. Garcı́a-Magariño. Internet of things for
healthcare using effects of mobile computing: a systematic literature review.
Wireless Communications and Mobile Computing, 2019, 2019.
[20] B. Farahani, F. Firouzi, V. Chang, M. Badaroglu, N. Constant, and K.
Mankodiya. Towards fog-driven IoT eHealth: promises and challenges of
IoT in medicine and healthcare. Future Generation Computer Systems, 78:
659–676, 2018.
 IoT-based biomedical healthcare approach 53

[21] H. Ahmadi, G. Arji, L. Shahmoradi, R. Safdari, M. Nilashi, and M. Alizadeh.

The application of internet of things in healthcare: a systematic literature
review and classification. Universal Access in the Information Society, 18
(4): 837–869, 2019.
[22] S.M.R. Islam, D. Kwak, M.D.H. Kabir, M. Hossain, and K.-S. Kwak. The
internet of things for health care: a comprehensive survey. IEEE Access, 3:
678–708, 2015.
[23] N. Scarpato, A. Pieroni, L.D. Nunzio, and F. Fallucchi. E-health-IoT uni-
verse: a review. Management, 21(44): 46, 2017.
[24] D.V. Dimitrov. Medical internet of things and big data in healthcare. Health-
Care Informatics Research, 22(3): 156, 2016.
[25] S.B. Baker, W. Xiang, and I. Atkinson. Internet of things for smart health-
care: technologies, challenges, and opportunities. IEEE Access, 5: 26521–
26544, 2017.
[26] J. Qi, P. Yang, G. Min, O. Amft, F. Dong, and L. Xu. Advanced internet of
things for personalised healthcare systems: a survey. Pervasive and Mobile
Computing, 41: 132–149, 2017.
[27] F.T. Al-Dhief, N.M.A. Latiff, N.N.N.A. Malik, et al. A survey of voice
pathology surveillance systems based on internet of things and machine
learning algorithms. IEEE Access, 8: 64514– 64533, 2020.
[28] M. Elkhodr, S. Shahrestani, and H. Cheung. Internet of things applications:
current and future development. In Innovative Research and Applications in
Next-Generation High Performance Computing, pages 397–427. IGI Global,
2016.
[29] N. Boudargham, J.B. Abdo, J. Demerjian, C. Guyeux, and A. Makhoul.
Investigating low level protocols for wireless body sensor networks. In 2016
IEEE/ACS 13th International Conference of Computer Systems and
Applications (AICCSA), pages 1–6. IEEE, 2016.
[30] P. Yang, J. Qi, G. Min, and L. Xu. Advanced internet of things for perso-
nalised healthcare system: a survey. Pervasive and Mobile Computing, 41:
132–149, 2017.
 Chapter 4
Impact of world pandemic “COVID-19” and an
assessment of world health management and
economics
Hemanta Kumar Bhuyan1 and Subhendu Kumar Pani2

For the COVID-19 pandemic, international health facilities have been issued in
financial section with difficulty. According to the economic status, several hospi-
tals and healthcare facilities loses a month in sales. Furthermore, providing an
adequate healthcare solution to COVID-19 could cost $52 billion (around $8.60 per
person) for low- and middle-income countries (LMICs). Could burden have a sig-
nificant effect on health treatment, surgeries, and outcomes through the use of
COVID-19. This year the World Bank predicts that the global economy would
contract by about 8%, with developing countries bearing the brunt of the burden.
Lack of planning was a key in dealing with healthcare facility issues everywhere.
On a national scale, items such as gowns, gloves, facemasks, syringes, disin-
fectants, and toilet paper ran out. Healthcare worldwide has felt threatened by
COVID-19’s findings and has responded by formulating new programs to deal with
pandemics. In this article, we will talk about the financial implications of COVID-
19 that include clinics, surgery, and medical procedures on both the US and foreign
healthcare systems, in the United States and abroad. In the case of natural or
human-made disaster, the United States and the rest of the world would make profit
from being prepared for a single blueprint to react.
By December 29, 2019, the novel coronavirus (nCoV) sparked a worldwide
pandemic. Such groups—especially the older adults and those with weakened
immune systems, such as low blood pressure, cardiovascular disease (CVD), or
asthma, or chronic kidney disease—are at an increased risk of the infection.
Systemic socioeconomic status (SES) is linked to a higher prevalence of diabetes,
hypertension, coronary artery disease, and obesity in both demographic groups.
Those already established conditions increase the likelihood of catching the cor-
onavirus (2019-nSRS-2), also known as severe acute respiratory syndrome cor-
onavirus (SARS-CoV-2). Emphasis should be put on the root of the issue at

1
Department of Information Technology, VFSTRU, Guntur, India
2
Krupajal Engineering College, Odisha, India
56 The Internet of medical things

vulnerable periods. The study of the social determinants of health (SDOH), which
involves research on the various facets of a population’s health in the context of
various aspects of their society, will assist policymakers in managing health crises
in ways that treat everybody fairly. During the COVID-19 pandemic, the SDOH
directly affected patient recovery and health outcomes.
Determine how these three variables influence national healthcare, social,
and economic well-being. Many forms of quantitative and nonnumerical data
have been employed. We have found small well-equipped clinics, poor rule
awareness, unemployment, and shortage of research equipment to be influential
in this COVID-19 infection. To combat the spread of diseases, enforcement
policies that are tough yet respect people’s rights could be more effective. The
delivery of adequate medical treatment is necessary for setting up appropriate
medical facilities.

4.1 Introduction

Once there were 7.5 billion inhabitants on Earth, 1 year ago, no one could have
expected COVID-19. World Health Organization (WHO) was the first to identify a
COVID-19 on December 30, 2019 and put the disease on a global emergency list
on January 30, 2020. But also scientists say that the virus was at work for some
time before this. Respiratory problems may occur with single-stranded RNA viru-
ses [1] in the air. It is estimated by the CDC project an infection, or virulence (R0),
of 2.5 for this new Corona virus strain. It found one hundredfold greater lethality of
the mutant variants than of the normal influenza virus [2,3]. When the pandemic
was already new, the global production chain was disrupted.*
About 51% of Americans recorded a loss of income and an increase in
unemployment of 14.7% due to the outbreak [4]. It has been difficult for young
people to make their mortgage and cover the bills this year due to missed income
and jobs. These consequences are more likely to occur in young adults in
America, who have not saved enough for an economic recession. Predictions
show that one in five tenants was likely to face homelessness this year. COVID-19 has
foreign implications as well. As soon as the Chinese government imposed a
mandatory quarantine, several countries lost a large portion of their critical trade
products. It is increasingly clear that many countries, particularly those in the
developing world, heavily rely on China. Unfortunately, several valuable equip-
ment pieces, such as respirators, gloves, were lost in the explosion. In turn, the
lack of appropriate personal protective equipment (PPE) promoted the global
pandemic. The World Bank expects global growth to drop by almost 8% in places
that have seen the most significant losses in development and 2 trillion dollars
this year [5]. COVID-19 has had an influence on all US healthcare services and
foreign organizations as well.

*
Projected to be the leading cause of USS disease costs in 2020. It is estimated that it would cost the US
economy $3.3 trillion. It is forecast to cost the US$3 trillion in the next 2 years [3].
 Impact of world pandemic “COVID-19” 57

4.2 Impact on societies

According to the National Collaborating Centre for Health, discriminated popula-
tions are classified as “Groups and societies who are discriminated against (based on
socioeconomic and political status), as well as socially and economically dis-
advantaged, because of unequal power relationships.” This time, this definition of
those who are un- or underprivileged has never been more on target [6]. The pan-
demic affected poor and underprivileged populations in the US anti-Hispanic, anti-
African American, and anti-working class. Despite having poor wages, low-income
people have been shown to be greatly impacted by COVID-19 [7]. Any businesses
have felt the consequences; nevertheless, personal utilities, bars and restaurants,
banking, and manufacturing remain critical in the overall economic situation. This
broad sector is made up of minorities such as African-Americans, Latino-Americans,
and women, which is especially important for developing innovative ideas. It has had
a disproportionate effect on minorities and women, leading to an increase in pre-
judice and exclusion. The statistical fact cannot discount the importance of knowl-
edge of working with populations that are disproportionately composed of members
of Latin and African-Americans that their jobs and income levels are lower than
Caucasians’. The poor are far more likely to be affected by the economic effects of
COVID-19 than the middle class or the wealthy. The onset of COVID-19 also raised
concerns about race relations between Asian-Americans and the police. This pre-
judice study has generated an astounding 1,497 complaints now, according to the
Asian Pacific American Center, which notes that a good number of these complaints
originate from only the first 8 weeks of operation. To the tune of 50%, about 50% of
these cases originated in two states—California and New York. Moreover, IPSOS
conducted a poll in the United States, which showed that approximately 30% of
people there attributed the COVID-19 pandemic to China. Some further findings
discovered that almost 30% of participants have seen others accuse COVID-19 for
Asian crime. These results are worrying since they show the pandemic is strength-
ening existing patterns of prejudice against the community who faced it before.
As a result of Creativeþ, there have been significant impacts on women. There
are many more ways for women to be engaged in work, home, and household
relative to men. The COVID-19 epidemic has clearly compromised this middle
ground. Schools and daycare facilities are closing due to the virus, so families
struggle to manage their babies. For home-based working mothers, this will con-
tribute to their daily tensions. Since the discrepancy in compensation between the
genders is such, she can simply scale back her job commitments to accommodate
the existing family obligations and concentrate on the important ones. As a result of
being culturally removed from the office, female employees of the service sector
may be significantly impacted. Many women with an essential function outside the
home return to their jobs even if they might have difficulty finding childcare.
Societal and occupational responsibilities have increased, leading to women bear-
ing an extra burden in the workforce and the home, which may lead to women
having a more significant stress load than previous pandemics do. The so-called
COVID-19 pandemic has hit the lower income strata of the community as well.
58 The Internet of medical things

Another is the workers of supermarkets. While being deemed an invaluable facility,

they make only $11.37 every hour and provide minimal facilities and few to no
healthcare coverage. A significant proportion of the Latino workforce is working in
supermarkets, and thus COVID-19 has heavily affected the Latino culture. This is a
bad state of affairs for certain retail employees who would put their well-being at
risk to obtain modest or insufficient compensation. This is, of course, that people
who serve at low-paying service positions in grocery stores must still be supported
because it is their lot in life.

4.2.1 Impact of social manner on COVID-19

The release of the COVID-19 pandemic has created a substantial psychosocial gap
in the American population. As a result, it has created alarm, trepidation, concern,
uneasiness, and mass disorder. The impacts COVID-19 has on the United States
would be discussed in this segment [8,9]. After the development of COVID-19,
Americans have suffered a detrimental mental impact. After the start of the plague,
the various human mental conditions such as anxiety and depression have
increased. At least 40.9% of participants experienced at least one psychiatric dis-
order, according to a CDC study. These involve signs of drug dependence,
depression, post-traumatic stress, and paranoia, all of the anxiety. And 50, and
about one-quarter of participants in the study who expressed thoughts of killing
themselves or attempting to kill themselves were in that age group. This study also
discovered disparate psychological effects on African-Americans, Hispanic indi-
viduals, unpaid domestic service employees, and critical workers concerned about
their mental health. According to the report, all three factors rose in June of 2020 by
comparison to the same month of the previous year.
Due to the stigma of COVID-19, people will experience significant psycho-
social impacts similar to those related to the SARS epidemic of 2003. The health-
care professionals dealing with patients who had SARS became more stigmatized.
In the same manner, healthcare staff caring for patients who are infected with
COVID-19 can often suffer from this occurrence. Stigmatization can incite victims
to think that they are being discriminated against and perceive others to be fearful
of them, to perpetuate their powerlessness, and reduce their standing in society,
which can contribute to feelings of being powerless and social alienation. This is
so, because stigmatized communities can trigger unhealthy behavior, which can
increase the spread of disease. Healthcare workers face more social marginalization
and personal vulnerability than financial risk.
The first widely acknowledged approach to combat the spread of COVID-19 is
quarantined. A vaccination is an essential part of managing this infection, although
there are potential problems. It has unintended detrimental societal consequences
due to being closed off and quarantined. The critical factors that aggravate the
condition include money, lack of necessities, and the possibility of a deterioration
of false belief about the disease reports. It could result in signs that are close to
obsessive–compulsive anxiety as it goes on to cause further loneliness. Repeated
washing and testing of temperatures cause maladaptive or aberrant habits, such as
 Impact of world pandemic “COVID-19” 59

compulsive cleaning or checking. Elderly patients and healthcare providers are at a

risk of long-term mental deterioration because of this pandemic. Several factors
contribute to this fact: carrying a great COVID-19 count, the possibility of infec-
tion, lack of PPE, new and excessive visibility, becoming a big part of moral and
ethical decision-making, and working overtime Additional caregivers must further
alienate themselves from relatives as a result of this. Anxiety and depression may
occur as a result of these stressors. If clinicians make an effort to look at clinical
factors and behavioral ones, this provides a psychological evaluation for addiction
and drug misuse. This is very problematic when healthcare workers put their well-
being at risk as they handle others. Even then, we would further open a dialog on
the mental well-being of employees at the frontline workers [10].
Facebook and other social networking have amplified the psychosocial pro-
blems COVID-19 generated. Soon after the advent of COVID-19, confusion and
fear-mongering escalated on social media. Social network rumor, paranoia, and
hysteria are directly proportional to each other regarding the number of platforms
available sources. Ironically, the effect of social media on the Chinese population
has become that of bigotry. As a result, outdated knowledge regarding eating pre-
ferences, social media biases, and misleading assumptions about the Chinese peo-
ple have emerged. The social media revolution has been mixed with disinformation
and has enabled the dissemination of bigotry and paranoia; as a result, not per-
formed properly, COVID-19 therapy may have unwanted side effects. Terror,
panic, anxiety may exacerbate these symptoms, and confusion. There must be
greater focus on dealing with the psychosocial issues associated with this drug,
particularly in fragile groups, including the elderly and people with preexisting
mental conditions.

4.2.2 Impact on healthcare facilities

COVID-19 hospitals had to build additional negative pressure spaces, recruit repla-
cement personnel, pay overtime for employees, and labor to train them, all of which
necessitated additional money and contribute to the existing scarcity of PPE. To free
up medical workers’ time and beds, both nonurgent and elective surgeries were
halted. As a result, there was an approximately 90% reduction in-office visits, and
almost all outpatient care was provided by video-chat services, which may have
contributed to a drop in patient anxiety. Due to increases in COVID-19 costs and the
loss of outpatient visits, several hospitals were placed under financial stress [11].
It was especially difficult for academic medical centers in the United States.
During the last few decades of their growth, they have increased their debt sig-
nificantly as they have undertaken numerous renovations and additions to the med-
ical facilities. Although running on tight margins, academic medical institutions also
functioned as a safety net, becoming one of the most costly ones in the United States.
Excessive pre-pensionary solvency, compounded by excessive therapeutic revenue,
jeopardized these organizations’ capacity to cope with COVID-19.
Veterans’ Administration hospitals were often dealing with much of the same
problems; they could, since they operated as a heavily centralized, nonprofit hospital,
60 The Internet of medical things

draw on numerous revenue sources, which was not as bad for them. For certain
physicians who practice in the Department of Veterans Affairs (VA) specialty and
procedural practices, the income impact of cancelled treatments was not as worri-
some since they do not get money for their services the same way fee-for-based
organizations do in the private sector does. Often, as a VA system works as a whole,
it will change the supply chain to serve best regions devastated by the pandemic. The
Department of Veterans Affairs provided over 16,500 intensive care units and offered
3,000 ventilators and 4,000 medical assistances for deployment in the form of
deployable disaster services. The Veterans’ Administration switched several of their
official trips to telehealth care. It was done by the VA through the usage of telehealth,
employing the first Chief Telehealth Officer in 1999. Before the arrival of the pan-
demic, the VA predicted the rapid increase in virtual treatment. Prompted by the first
COVID-19 cases in the United States, they started to invest in the technology,
including providing patients with mobile computers, pulse oximeters, and cell phones
to track their temperature remotely, such as smartphones.
During the plague, there was a campaign to help the elderly of our community.
One small nursing home outbreak in Washington State helped start the pandemic. A
nurse at the nursing home developed an infection called COVID-19 on February 28.
Both employees and residents in the facility reported signs of respiratory disease on
that day. By March 18, a total of 167 COVID-19-infected people, 50 COVID-19
travelers, and a lone agent had been found. Accordingly, the hospitalization rate was
high for 54% of the inhabitants and 33% of the workers, resulting in a fatality rate of
33% [12]. More than 40% of all pandemic patients and staff in long-term care
facilities have succumbed to COVID-19. We found that eight out of every ten cancer
of the COVID-19 population was in people aged 65 and over. Thus, the Coronavirus
Assistance, Relief, and Economic Security Act allotted $5 billion to long-term care
centers and older people’s homes for the elderly. Thanks to the improved funds,
additional infection prevention procedures, as well as personnel and program being
put in place for those unable to leave their house, the nursing home was able to have
extended visitation features to its residents [13].

4.2.3 Worldwide impact

4.2.3.1 China
According to the initial reports in Wuhan in December of 2019, China was the
epicenter of the pandemic. Within a month, the Hubei provincial government in
China had enforced a state of emergency on Wuhan. Buses, trains, subways, and
water taxis ground to a halt: several vehicles were unable to navigate inside the
approximately 9 million inhabitants of the region. The airports and rail stations were
shut down for the duration of the strike. Restrictions on intercity and domestic travel
were enforced elsewhere, though unfortunately, these preventative steps limited
disseminating the virus. The country had a GDP growth of 6.8% in the first quarter of
the year, nearly after 30 years. A fast lockout led to the disruption of many citizens’
lives. A great fascination is expressed in the West as China lifted the last of its
controls in April of 2020, and the rest of the world awaits its return to normalcy [14].
 Impact of world pandemic “COVID-19” 61

4.2.3.2 India
It was a step in the right direction for the Indians, who, after ages of isolation and
British colonial domination, had finally regained their freedom and identity, had
the opportunity to start constructing a modern democratic nation on the Indian
model. Trissur is the first in India to be hit by COVID-19, on January 30, 2020. The
government initially employed policies to limit the spread, including quarantine of
persons who traveled from high-risk locations, tracing those who had contracted the
virus, and preventing movement of those with a large caseload of cases. When the
number of cases grew, local dissemination became apparent, attention turned to
dealing with the virus, and control steps began to dominate the Ebola. The moves
India has introduced are similar to those in China, including closing public venues
and rerouting foreign air transport. Furthermore, these regulations inflicted a
financial burden on the local economies and had long-term detrimental effects in
the agriculture, industrial, and utility sectors [15,16]. The disease spread among the
countries with which India had trade relations and within them, which put a con-
siderable dent in India’s exports. Furthermore, the general pandemic and resulting
mass incarceration have significantly increased anxiety and depression in India.

4.2.3.3 Brazil
COVID-19 did not get to Brazil until February of 2020. Still, it soon became a
celebrity in this field as a hotbed of creativity contrast to some national responses.
State/Federal governments initiated measures. President Bolsonaro attracted wide
backlash for his apparent inattention to the virus. Since he spoke out so passionately
about FU lockdowns, he usually touted the overhyped hydroxychloroquine cure and
had a generally low attitude toward the pandemic. An estimated 13 million people are
cut off from access to safe water in Brazil’s informal slums, poverty-stricken shanty
towns. Due to the global spread of the pandemic, the economic consequences in Brazil
have been substantial. On March 9–13, the equity exchange in São Paulo fell by over
15% to its worst weekly percentage decline since the Great Recession of 2008.
Additionally, the GDP declined by 11.4% in the second quarter of 2020 relative to the
same period in the previous year. As of September of this year, with an incidence of
4 million and a mortality rate of about 6, behind just the United States and Brazil in
the world, the United States in incidence and rate of, the United States, Brazil has over
a million reported cases and over 125,000 people die per year from them [17].

4.2.3.4 Singapore
Singapore’s reaction to the pandemic has been extolled as one of the most pro-
mising ones in the world. With the first cases arising in December 2019, the nation
was one of the first to register COVID-19 and initially was second only to China in
overall cases. The SARS-CoV epidemic of 2002 revealed several vulnerabilities in
Singapore’s capacity to cope with pandemics and was the catalyst for the nation to
patch up those deficiencies. The country developed the National Centre for
Infectious Diseases (NCID) and 900 quick responses to public health preparedness
clinics. Daily drills were held to simulate the arrival of a pandemic. As a con-
sequence, Singapore was well prepared to conduct mass screening of possibly
62 The Internet of medical things

infectious persons. They have identified themselves as one of the most prolific in
research and touch tracing. The National University of Singapore (NUS) Yong Loo
Lin School of Medicine created an engaging series of comics entitled “COVID-19
Chronicles” that provides essential information about the virus in a format that is
easily understood by most demographics. Messages such as these have been widely
disseminated across the world and have received universal acclaim.
Additionally, Singapore used several strategies similar to other countries that
suffered outbreaks, such as reducing the scale of meetings, promoting physical
distancing, restricting movement, and screening and quarantine anyone entering the
nation. Strict fines were levied for breaching these rules. Interestingly, Singapore
did not initially resort to issuing a systematic shutdown or shutting schools until a
flare-up of cases leads to a 21-day order on April 28, 2020. This was some months
after the virus was first observed in the world [18]. Despite the country’s progress
in managing the contagious burden of the epidemic, owing to the interconnected-
ness of the global economy, Singapore have not sparred the economic effect of the
pandemic as the country recorded 13.2% contraction in GDP in the second quarter
of 2020 as opposed to the previous year [19].

4.2.4 Impact on low- and moderate-income countries

How long before a pandemic breaks out in the high- and low-income countries?
Health problems that have been caused by long-standing barriers such as inade-
quate healthcare coverage, lengthy insecurity, limited clean water access, and
highly crowded slums are exacerbated by the appearance of the outbreak of
COVID-19. Realities such as mutual distancing and constant handwashing and
mass monitoring render the deployment of that almost unlikely. Moreover, in
LMICs, people cannot afford to skip a single day of operation, and the economies
are less capable of taking a full shutdown [20].
The concept is generally applied to cases when the United States or China does
something, but some countries bear the flu. Indeed, the global consequences of
lockdowns and ensuing recessions have trickled down to the lower middle econo-
mies. However, between April and May of this year, remittance receipts in
Bangladesh dropped by 19% and 41%, respectively, in the Kyrgyz Republic.
Remittances are a significant component of GDP in lower middle–middle-income
countries, and many families depend on them for their survival. Pacing themselves
would be tough, not to say the least, as the faltering global economy adds fuel to the
virus. Fortunately, the majority of LMICs responded quickly by implementing
school closures, limitations on public meetings, and travel was able to slow the
outbreak’s spread. Moreover, in the global community, there is an effort to help
these countries meet their needs [20,21].

4.2.5 Impact on international healthcare facilities based on

medical services
COVID-19 has affected almost every facet of the world’s markets, not least of
which was healthcare, with staggering problems in treating the global pandemic.
 Impact of world pandemic “COVID-19” 63

Unpreparedness was a big part of the problem of healthcare services. PPE [22] was
also lacking for healthcare staff. Different healthcare staff used PPE to protect their
health during medical services. The Pakistani analysis discovered that just 34.5%
had helmets, and 0.5% had face masks or goggles. In Jordan, a survey claims that
just 18.5% of physicians had access to PPE. Even the US healthcare sector was
troubled by the accessibility of prescription supplies. Approximately 15% of the
doctors indicated that they did not have N95 respirators. Approximately 50% of the
doctors reported that they did not have facemasks, and approximately 50% reported
that they did not have complete suits or gowns. Furthermore, about 7% of physi-
cians recorded caring for patients who had contracted COVID-19, and approxi-
mately 80% of PPE had been reused. Healthcare services across the world
requested nonmedical PPE donations and resourceful residents modified products
to meet this need. People around the globe faced similar issues with the availability
of intensive care units and ventilators. We did not have the resources to test locally.
Disturbances in global supply chains caused more shortages of food and
consumer goods.
The pandemic has diminished efforts for the diagnosis and treatment of non-
COVID-19 disease as a side effect. Noncommunicable disease (NCD) preventive and
recovery interventions have been compromised since the COVID-19 pandemic
started. As the virus advanced, healthcare staff, who had usually handled non-
catasthma cardiovascular conditions were pulled from treating patients to assist with
the COVID-19 response. In compliance with public health recommendations, certain
procedures and appointments were also delayed. Shrinking of public transportation
impacted patient’s travel times to appointments. Sometimes, cancer and diabetic
patients were unable to get the treatment and medication they needed. Ironically,
these results were seen more in low-income nations, where countries still have a large
percentage of their national income tied up in battling the pandemic. The results of a
survey in Bangladesh, Kenya, Pakistan, and Nigeria have found decreased access to
antenatal and vector control systems, as well as tuberculosis and HIV programs,
before the pandemic. Moreover, there was a notable decline in family income at the
same period as healthcare expense rose [23]. COVID-19 also prompted healthcare
providers to develop contingency plans that guarantee that necessary treatment is not
postponed or denied due to a global viral or another epidemic. Alternative solutions
such as telemedicine have gained a significant foothold in healthcare settings
worldwide and have aided in mitigating the impact of the epidemic on people.

4.2.6 Emergency caring

In several nations, the COVID-19 pandemic had a significant effect on the global
economy, so in turn, the Canadian Cardiovascular Society issued recommendations
to help cardiac surgery hospitals stay profitable. They call these recommendations a
“ramp-up” that meets all local health mandates while increasing elective cardiac
surgeries at a healthy pace. Experts think that normalcy of reduced catastrophe loss
can help limit the economic impact of COVID on cardiac surgery. From the
viewpoint of healthcare, a team from Germany performed a pan-European study on
64 The Internet of medical things

these two specialty practices that asked “excessive financial challenges due to the
trauma pandemic.” Ironically, being self-employed was shown to be a strong
indicator of financial hardship, one of which was referred to as “deplete” in the
study’s multiple regression model. Thus, according to their multiple regression
analysis, self-employed trauma and orthopedic surgeons should be supported more.
Like their cardiac surgeons, this profession shares a similar viewpoint that the
pandemic’s impact on the cost of their operations stopped unnecessary procedures.
The specialized trauma and orthopedic surgeons are unmoved by financial con-
siderations. Results indicated that nearly one-third of the group would give up
practicing because of the pandemic, and nearly two-thirds might benefit from
increased healthcare funding.
On the other hand, telemedicine offers a silver lining, with 44% of participants
believing that it will rise in value in the future. While the survey was established in
rather extraordinary conditions, a lot of open questions remain about its psychometric
testing. In 1 month at his 22-year-old private practice, there was a 29% reduction of
money on every occasion due to the worldwide epidemic. He lost money at a private
practice which he started in Italy before the pandemic 2 years before it struck
entirely. According to the authors of this chapter, all surgical professionals in all
countries who want to limit their career to just cosmetic surgery had done so due to
the lack of options due to financial difficulty. Overall, it describes the first instance of
a partnership between Latin American and worldwide healthcare stakeholders to
show the substantial contribution of COVID-19 healthcare resources to the medical
service economy [24]. Finally, there is a discussion of COVID’s economic effect on
the Latin American radiation oncology field. The survey conducted in Latin-
American countries like Argentina, Costa Rica, and Peru (the three countries had
fewer than 3% of the total of radiation oncology treatment centers) showed that both
had seen a 20% revenue loss from closure inactivity. Less than 3% of the total
radiation oncology facilities were still operational during the pandemic. They admi-
nistered the ASTRO survey in Latin America to view the pandemic’s overall finan-
cial effect on an underutilized field of healthcare. In Latin American radiation
oncology, use of telemedicine as a tool of the future was predicted.

4.2.7 Constructive response reduction strategies

Preventing the spread of the nCoV SARS-2 would entail a device protection solution
that considers viral potencies, disinfection tactics, environmental biobrave ability, and
human contact/risk reduction. Early symptoms diversely excrete the virus in several
layers and particles. Viruses are found in raw substance droplets and aerosols that are
found in blood and other body fluids. When a small number of viral mixtures
are generated and their density and speed vary, their traveling distance is affected.
Fine dust seems to collect around surrounding items or people. A more delicate, viral-
laden substance circulates in ventilation patterns. Virulence is based on the quantity of
an encapsulating organic environment. Second, try and limit the release of hazardous
coexisting microorganisms to work areas, breaks, and lavatories. To help avoid the
spread of illness, hold sick and possibly infectious patients at home is essential to
 Impact of world pandemic “COVID-19” 65

minimize the incidence of disease. Whenever they wear masks, risk to their coworkers
decreases. Since masks avoid involuntary hand and mouth gestures, they are helpful in
terms of personal well-being. The regular use of caps, gloves, mask, and glasses to
reduce the spreading of disease. The vaccines must be available to everyone, espe-
cially those at a risk of COVID-19 infections must be easily accessible.
Antimicrobial gels from the Centers for Disease Control may be made non-
contagious by primary low-level disinfectants. Usually, they are between 50%
alcohol and 60% gel solutions. This agent protects the alcohol from bacterial eva-
poration; therefore, there is plenty of time for the virus particles to come in touch
with it. There are significant considerations that have to be considered: focus and
touch time. Alcohol in water just reduces the total alcohol content gradually, as it
evaporates rapidly, which prevents more extended touch periods. An intact skin
barrier cannot keep viruses out of the mucus membranes but can provide defense
from transferring to them. Thorough handwashing of soap and water at the start of
the day reduces the infective bioburden brought in from the outside from interac-
tion with contaminated surfaces and suspended droplets and aerosols. Daily hand
hygiene after interaction with all aspects of the patient’s treatment to decrease the
risk of disease transmission can necessitate special hand wipes. One may minimize
the chance of being sick by washing hands, arms, face, or spitting in that manner.
Because it is safer for staff, it also helps prevent the movement of infective bacteria
to workspaces such as desks, computers, mice, and touch screens. The surface of
the tools, including drawer handles and bars, should be disinfected according to the
number of employees who use them. Plastic enclosures are more efficient at
keeping liquids away from compromising electronic connections through main-
taining color-coded containers and gel dispensers in work environments; con-
formity with safety disinfection procedures is supported by visual cues.
Mechanical ventilation controls lower the particle level of HVAC (reduces the
concentration of bacteria in the airflow). The usage of higher airflow speeds (above
612 exchanges/hour) at sites over 612 volumes helps remove infectious aerosols. Still,
it prevents desiccate or sedimentation of virus-laden or other aerosol particles. When
working in the ventilation and HVAC environments, work only in the fresh air inflow
zones and stay away from the return air inflow zones. Break rooms offer unique
obstacles. Lower volumetric airflow has the effect of making aerosols more persistent.
During meals, masks are off. This makes it easier for the droplets and particles to be
dispersed, facilitating the spreading of the infection(s). The locations that incorporate
some high-frequency touch surfaces, such as the door handles, the coffee cup, and the
tabletops. Cultural considerations should take precedence. Fatigue leads to a loss of the
ability to do repetitive and complex activities. It degrades for many hours; however,
fully after 12–16 h. The standards for infection prevention and prescribed practices are
likely to become less stringent as an individual’s level of exhaustion increases. There is
a greater likelihood of infection because of illness among people who live and function
close to each other. Controls on productivity are enormous. When job tasks are made to
anyone at random, the spread of personal creative ideas is maximized. Studies done by
Mascha et al. showed that consistency in team tasks would help minimize staff cross-
immunity. They recommend 1 week of work and 1 week of vacation for each
66 The Internet of medical things

consecutive week to help employees who are overburdened or highly stressed.

Prevention of burnout is often done by recognizing the style of the leader and planning
for leadership to prevent burnout. Empathy and influence over others are associated
with increased professional achievement [25]. Other job habits involve wearing high-
quality facemasks, distancing oneself, testing the community for potential damage after
exposure, and using water decontamination within a reasonable period, with well-
thought-out cleaning protocols in place.
Training aspects involve wearing top-quality gloves, social distancing oneself as
far as possible, measure oneself in the community, and thoroughly disinfect for a
decent period. Not performed correctly, COVID therapy may have unwanted side
effects. These symptoms may be exacerbated by terror, panic, anxiety, and confu-
sion. An alternate strategy has been used in many healthcare institutions, such as
using telemedicine, proving helpful in preventing the global spread of NCDs.
Compliance with infection control and suggested practices can rise with wear levels.
Telemedicine’s merit lies in finding a means of minimizing infection while sacrifi-
cing overall standards. COVID-19 has made operational shortcomings for healthcare
providers to provide fewer effective services for patients through medical procedures.

4.3 Impact of health on the emerging COVID-19

pandemic

People started being sick as the nCoV (Chinese New Year, January 2020) traveled
quickly through China. On December 29, 2019, the first four nCoVs were
announced. Wuhan is the biggest city in Hubei province, which is located in the
middle of central China. The symptom was strange pneumonia of no apparent
cause [26]. Early evidence shows the business touch. More and later, it was dis-
covered that disease could be spread from person to person, often by human touch.
The specific illness triggered by 2020-NP was called by the WHO on February 11,
naming it COVID-19 on the follow-up date [27]. People with preexisting asthma,
CVD, hypertension, chronic kidney disorder, and/an obesity are particularly vul-
nerable to the findings of the CDC proved that all who have these conditions have a
higher risk of serious illness from COVID-19. Although SES may increase the risk
of CVD, particularly among those who have asthma or hypertension and many
others, which are also risk factors of CKD, including depression, diabetes, and
obesity (SES). CDC estimates that 94% of COVID-19-treated patients succumbed
[28]. These particular circumstances increase the likelihood of SARS-2 infection,
and as a result, people in these areas must have the support and services to survive.
This study explains the importance of SDOH during pandemic (COH). Precausal
precautions should be taken to keep the transmission within abounds.

4.3.1 Social determinants of health

The SDOH encompasses five main social and economic conditions outlined in
Figure 4.1 improving access to all resources, improving the social and community
resources around our welfare, the economic conditions around our communities,
 Impact of world pandemic “COVID-19” 67

Health and
healthcare

Social and
Economic Social community
stability determinants of context
health

Neighborhood and
Education built
environment

Figure 4.1 Different domains of social determinants of health (SDOH)

and enhancing our built environment. Access to affordable services, having

healthcare, having primary healthcare, and understanding health needs are all
components of health and fitness, and well-being. Have a low level of health lit-
eracy will hinder patients in finding their way around the confusing health system
and take away their comprehension of medical advice or medications. People who
have no health benefits would be less able to use or still have regular medical
coverage, which may complicate CVD, asthma, and cancer management. Context
is just as important as it is in how an individual leads his or her life as it is in
defining what he or she does. The practice of group engagement and inequality falls
within this domain. Religious and group connectedness reduces the risk of pre-
mature death. The built environment encompasses infrastructure, travel, food, air,
and environmental issues [29]. Secondary causes of asthma have been shown to be
related to air quality. The CDC also discovered that people with asthma are at an
increased risk of full-blown disease to occur [30]. Safety is paramount in fitness.
Residents are more inclined to stroll or run in their communities if they are sure that
their neighbors are doing the same. The concern about danger and security makes
for mental well-being as well. Activity impairs the immune system. This has been
seen in previous studies to be the case for poor people and those in disadvantaged
communities. These neighborhoods also experience stressful conditions due to the
additional stressors of “overcrowding, high violence, inadequate facilities, and poor
social support.” The schooling process encompasses high school graduation,
registration in college, and fluency in a foreign language.
The more educated the one is, the longer his/her lifespan. It is often necessary
to provide health records in a helpful way to the customer, depending on their
degree of schooling. Economic well-beings are income, employment, household
68 The Internet of medical things

structure, health, education, housing, food, general safety, recreation, and nutrition
(SDOH). In economic hardship, the number of people who have activity-impairing
chronic conditions goes up. Many people become unemployed and are afflicted by
homelessness, drug misuse, or become sick due to the effects of unemployment. An
explanation of SDOH is that there are financial resources, such as a salary, life
experience, and various forms of social help. Simply placed, in short, you are the
way you were born, learn to be, that is, the way you live and function your whole
life [31]. They see the entire guy. These various diseases have a widespread effect
on the well-being of people and the welfare of whole populations. Because of these
socioeconomic differences, SES can be equated with social inequality (SES). The
better on the socioeconomic scale, the worse the health consequences. Lifespan is
shortened for people on the bottom half of the social scale. If you look at things
socioeconomically, health problems build on the inequities already befalling a
community section.
Many characteristics are interrelated and play a major function in the eruption
of the COVID-19. For example, an individual’s education affects his or her eco-
nomic stability and what neighborhood he or she is eligible to live in, which affects
his or her social well-being. Consequently, socioeconomic variables are likely to be
important in infection and mortality At the time of data collection, the researchers
discovered certain counties in New York, including the Bronx, Brooklyn, and
Queens, which also had a significantly higher death rate among people of low SES.
This case is another example to consider when considering a child growing up in a
low-income home. It is difficult for low-income families to provide for their chil-
dren, which could lead to low-quality schools.
There would be no equality in education, no matter where this kid resides, as
long as it is different from a wealthier suburb or a more decadent school system
because government schools are financed by local, national, state, and federal
resources [32]. Any funding for our social projects will come from personal income
and property taxes. The affluent areas and districts would raise more money; there-
fore, they have more money to spend. Districts that serve low-income students get
smaller salaries and inadequate services, while poorly educated teachers receiving
small wages are included in low-income schools. It does not matter whether you get
average or below-average grades as well as long as you plan on pursuing an above-
average academic career in college. And if it is from a low-income family, the kid
would always have a few high-paying career prospects. This places the infant in the
same financial situation as his or her parents. People who live in poverty-ridden
communities have higher chances of remaining that way. The five factors can be seen
as a kind of a cycle that affects an ever-changing scenario, even today.

4.4 Health in the clinical system

4.4.1 Clinical knowledge

The need to access, store, interpret, and understand essential clinical knowledge
make health decisions [33]. Being educated about one’s disease will also
 Impact of world pandemic “COVID-19” 69

include reading and comprehending clinical information brochures, health

advice from a doctor, and patient. It is challenging to take care of one’s health
even though you can read health-related content because of a lack of literacy. In
general, individuals with lower levels of literacy have worse health results.
Sometimes, studies have shown some demographic classes to have lower levels
of health literacy. Disadvantaged, uneducated, and minority populations are
more likely to have poor health literacy [33]. Patients that show evidence of
having poor literacy may have high overall literacy and good verbal fluency.
Health awareness is critical in a public health emergency, and it aids in the
control of COVID-19. Any understanding of infectious diseases, such as how
well viruses survive and spread in the environment, helps us consider the con-
ditions in this case. In terms of the future influenza C outbreak, understanding
health awareness can empower individuals to apply social distancing and other
steps to keep the virus at bay.
Social distancing can be crucial in reducing the COVID-19 in April and May
of the following year. Determined if each community with a certain degree of trust
adhered to a prescribed distance from other groups, 52% and 31% were “moder-
ately” “significantly” convinced in their conviction that social distancing them-
selves from others was successful in warding off illness during the upcoming
COVID-19 episode [34]. Of course, however, almost 14% of participants showed
doubt regarding social distance and its effectiveness in preserving life. Overall,
88% of the Americans who took part in the survey said they “still” or “very often”
distanced themselves from crowds. In light of recent developments, the percent
distribution dropped from “always” to “often” to “frequently” and in several cases
to “usually.” For those who “think it matters,” 79% exercised distancing, while just
half of those who “think it doesn’t matter” did. Freedom, 43% “most of the time”
did. According to these facts, health awareness was vital in determining whether a
person could grasp the seriousness of the COVID-19 pandemic and whether he
might meet safety guidelines, such as distancing himself from family members
during the flu, and whether or seek out loved ones.

4.4.2 Access to medical- and primary-care doctor

Admission to healthcare is better known as “timely delivery of personal health
benefits to get the best potential health results.” Many individuals encounter limiting
healthcare challenges, which can impact their well-being. An obstacle in healthcare
includes, but is not restricted to, such issues as a shortage of transportation for
medical services, low or no benefits, the lack of accessible healthcare, or limited
funding. At poverty levels and minorities, the lack of healthcare is pretty common.
More than a quarter of American adults indicated that they would avoid medical
treatment if they had a fever and dry cough [35]. Seizures and flu-like effects are
most often associated with this strain of smallpox. Answered, including when shown
video footage of themselves becoming poisoned with COVID-19. Many adults who
said they were unable to afford health insurance, ages 30–30 to 39 with annual
incomes of less than $40,000 and had just a high school diploma or less, were more
70 The Internet of medical things

likely to forgo seeing a doctor. In low-income people, it is linked to SES. Hispanic

and African Americans, and nonblack people have less access to healthcare. Primary
treatment may not be possible without health benefits. Or individuals might be less
likely to use healthcare services. This places the uninsured ones at risk of long-CVD,
hypertension, obesity, and diabetes screening out of those who do not have insurance
into question. To provide the provision of medical services, preventive insurance
must be made available. Minorities and the poor experience further hardships by
attempting to get quality healthcare. All of them are strongly supported by student-
based clinics. A student-run health center on the University of Nebraska Medical
Center campus, named the SHARING Clinic, serves as healthcare and healthcare for
communities in need and underprivileged in Omaha. As a result of the COVID-19
epidemic, this clinic is now closed. To the underserved communities, which might
still experience accessibility issues, these student-run clinics exacerbate the problems
they have because of it.

4.5 Role of health disease

4.5.1 Food deserts on cardiovascular disease (CVD)

A food desert is a low-income region where enough access to nutritious food is not
accessible [36]. A research discovered that people living in an environment with
little access to supermarkets and fruits and vegetables were at risk for CVDs in the
metropolitan area. They observed that income was significantly related to CVD
risk when it came to diet while also finding that healthful eating was not related to
CVD (18). Awareness that wealth has a more significant contribution to position
led them to equate low-income communities to individuals who earned low indi-
vidual income. The researchers discovered that an individual’s level of wealth was
more closely linked to CVD than their area income or access to food. Those indi-
viduals who had above-average gross household incomes in lower income areas
had lower CVD risk levels [36]. People with more money, who lived in poor
communities had healthier cardiovascular profiles. This suggests that, at least in
part, access to nutritious foods is not a factor that increases CVD risk. Furthermore,
this new research discovered that class is associated with increased mortality from
CVD, and lower SES [37]. This study shows an inverse relationship between lower
income and a greater chance of severe infection with CVD.

4.5.2 Role of food deserts on hypertension and chronic

kidney disease
Low wage is a known risk factor for both hypertension and CVD. Fruits and
vegetables prove to be more nutritious than they are less expensive. High food
prices restrict the consumption of nutritious foods by low-income households. In
low-income and minority areas, refined foods and fats and oils are favored instead
of fresh fruits and vegetables. Results: A research survey conducted by Suarez et al.
shows that 80.3% of the participants “still” or “much of the time” have fruits at
 Impact of world pandemic “COVID-19” 71

home [38]. This compares to 83% of participants who are in the top wage bracket
and who have access to grocery stores or farmers’ markets in their neighborhoods.
The families who lived in food deserts had more impact on their dietary choices,
blood pressure, and incidence of kidney disease than those who lived in other
locations [38]. It was discovered that people who lived in food deserts and had low
wages had lower serum concentrations of carotenoids. There are ways to get an
idea about how many fruits and vegetables you consume using carotenoids. It was
also discovered that individuals in food deserts have lower average protein,
potassium, sodium, and magnesium consumption but higher average wages. The
quantities of these minerals will reveal one’s acid load in the system. High pH
indicates an alkaline diet [39]. High protein foods (meat, dairy, etc., in particular)
trigger a rise in body acidity. Fruits and vegetables increase the number of organic
compounds in urine. When you have a diet of a lot of acidic foods, you are likely to
have metabolic acidosis and asthma, kidney disease, and other issue-related pro-
blems. Studies have shown that a heavy acid load contributes to obesity [40].

4.5.3 Role of SDOH on obesity

There are more fast food chains in food deserts than there are supermarkets.
Individuals who do not have easy access to affordable access to fresh, wholesome
foods have obesity [41]. According to the CDC, an individual is categorized as
“obese” as having a body mass index (BMI) greater than or equal to 40 [30]. Those
people living outside food deserts are more likely to be eating more fresh produce
than those who live in supermarkets. They are less likely to be overweight. The
likelihood of obesity being linked to a hypovolemic respiratory disease, recognized
as the Pickwickian syndrome, is greater among overweight individuals. It is unclear
whether this disorder is more common among people who are obese, although it is
theorized that they may have more difficulty taking deep breaths around the neck or
belly due to excess weight. Due to the buildup of carbon dioxide, it causes anoxia.
Adjoints may be secreted in reaction to respiratory difficulties [42]. BMI is the
body mass in kilograms divided by the square of the person’s height in meters. BMI
indicates when a person’s weight is within an acceptable range. People with a BMI
under 18.5 are undernourished. The BMI ranges between 18.5 and 25, which is
considered to be natural. A person has a BMI of 25 to 30 to 32, which places them
in the overweight category. A BMI of 30.0 or above indicates that a person is
overweight [43]. Throughout the study, it was discovered that 62% of the patients
died from viral infection. About 64% of nonobese people had to go into artificial
breathing, and of those who survived, 36% had to suffer from the infection. In
people with a larger proportion of body fat, the numbers of those needing artificial
breathing and deaths are notably more significant.
A BMI >40 was the second strongest independent predictor of hospitalization
in patients with COVID-19 at an academic hospital in New York City [44].
COVID-19 patients who required a ventilator were obese or morbidly obese,
according to French research. Due to decreased respiratory mechanics, respiratory
function, improved airway resistance, and diminished gas exchange, many patients
72 The Internet of medical things

will need ventilator support [45]. Even if some people who are obese, the breathing
muscles can be weaker because of fat on the neck or in the belly, as previously
discussed, and lung volume may be affected because of fat that restricts the ability
to expand the chest and take a full breath [42,45]. The findings also showed an
improvement in the risk of COVID-19 as BMI rose [45].

4.6 Social setting

4.6.1 Determination
People’s perceptions of unfairness and injustice can feed bigotry. Actions benefit
the well-off at the expense of the poor. In both the human and institutional levels,
healthcare levels, discrimination exists. Individual discrimination involves
encounters between patients and their healthcare professional attributable to eth-
nicity, gender, or some other discrimination that leads to a diminished ability to
access care. Healthcare services and the patient’s well-being may be impeded by
experiences that lead to illness. Hierarchical inequality can be seen in the context of
residential segregation along ethnic lines, gender lines, wage inequity, and so on.
The particular form of systemic discrimination (name, skin color, gender, sexual
orientation, national origin, socioeconomic class, age, religion, etc.) may impact
individual people as well as groups. Somehow, residential apartheid finds itself
contributing to the racial differences seen between African and Caucasian com-
munities. African-Americans live in concentrated poverty. Education systems in
these areas are inadequate, unemployment is very high, health insurance is una-
vailable, violence is a constant problem, and residents have no social support. Still,
life for the residents is complex [46]. This is because it is hard to maintain social
isolation in communities that are poor. Any relatives will not be able to keep their
distance from each other. In extreme poverty, minorities and blacks are more likely
to keep positions where it is not possible to telecommute [47]. This COVID-19
emergency has placed many minorities in the position of either needing to work and
putting food on the table, or remaining at home and keeping their families safe,
especially the Latino community because of a sheer number of jobs in the ware-
house work, manufacturing, maintenance, and custodial employment are common
among minorities [47]. While 35% of those infected with the virus do not identify
themselves as either race or ethnicity, it is readily apparent in the current statistics.
Many New York City residents have died from it than any other racial race [46]. In
New York, 29% of the population are made up of Latinos. Of all the COVID-19
people who die in New York, about 34% are people of color—22% of the area.
However, people of African descent in the United States have a 2.4-time greater
risk of death due to this infection than those of other races. The facts for the US
population, broken down by territory, are startling. According to the US Census
Bureau, African Americans are 13% of the overall population, although they have
suffered 32% of the COVID-19 deaths. On the other hand, white people are much
more likely to be at risk because they live in the United States than those in the rest
of the world [48].
 Impact of world pandemic “COVID-19” 73

4.6.2 Active part of the community

Whether you have social assistance, you are much better off psychologically.
Social connectedness, another definition for the term continuity, defines the kind of
relationships a group has with its participants. Thus, social capital, an index of
mutual wealth, perception of justice, solidarity, and perceptions of group benefit
are socially necessary [33]—directly or negatively for survival [49]. The more
unequal the society, the fewer people have access to social resources. More
research indicates that social connections, specifically, reduce both the income gap
and mortality. The more excellent social stability in a neighborhood, the lower the
levels of tension and anxiety and the higher the self-rated well-being for the resi-
dents. Immune related to the immune, cardiovascular, and neuroendocrine stress
may have significant effects on the body. People who are shown to have a greater
likelihood of developing atherosclerosis if provided with more excellent assistance
from others have been seen to have a reduced risk of CVD. As in California one
showed, social assistance serves as a bridge to social problems that Hispanics face
[50]. People and cultures have pulled together through this challenging period of
need. Clinical clerkships are placed on hold, which means medical students cannot
help everyone who comes to the office. Nationwide, medical students have stepped
up to support frontline doctors and even help out with infant care and pet care.
Nebraska Medical students can use the time they are afforded to take off their
clerkships to serve in the community because the supply of face and body protec-
tion equipment has been low among those who know how to sew because of
shortages of supplies for frontline personnel (PPE). COVID-19 is more risk than
other disease to survive with social stability.

4.6.3 Make surrounding foods

4.6.3.1 Access to healthy foods
Food is a must for existence. It is significant to one’s overall well-being. Those who
eat more fruits and vegetables have a reduced chance of serious illness. A balanced
diet consists of a wide variety of fruits, herbs, lean meats, legumes, soy foods, and
other sources of protein, as well as nondairy sources of fat. Poor dietary and health
problems, such as hypertension, diabetes, and cancer, have been shown to be
related to suboptimal feeding [51]. There are several elements in the built envir-
onment and community that influence each other. While healthier foods are usually
very accessible or in demand, they still face several obstacles in getting to the
public. Healing may also occur because people can quickly get to the right food, a
factor in the city and developed environment. A survey released in 2012 revealed
that the typical grocery stores are 2.19 miles away from households in the United
States [51]. To prevent anyone who does not have their cars or connections to
public transit from making a shopping trip, this has the potential to be a real
obstacle. Food deserts are regions that are found in low-income cities where people
have no access to nutritious food. Global crop production is estimated to increase
50% by 2025 [41]. Grocery shops are more prevalent in more residential areas than
commercial zones. Cheap food shops and supermarkets are chock full of toxic and
74 The Internet of medical things

unhealthful choices (higher saturated and trans-fat and higher calories). Individuals
who do not have ready access to fresh fruits and vegetables tend to have inadequate
diets. Because African-American and Latino neighborhoods are more likely to have
a higher concentration of fast food and grocery stores, they are also more likely to
be, on average to be less attractive to someone wishing to establish a new business
in them. However, the ethnic groups often may not have as good health results as
racial groups. Why? People living in food deserts have an increased risk of obesity,
which is mentioned elsewhere. Income is another essential factor of balanced food
consumption. There are solid and numerous studies that indicate people on low
incomes depend on lower nutrient-dense food. Processed goods tend to be more
costly, but they are not always unhealthy. People who cannot afford to buy fresh
foods purchase the packaged ones instead [51]. It is essential to consider food
deserts during a pandemic and that people do not have accessibility to nutritious
foods, particularly. When food is in short supply, it would be impossible for those
who do not have sufficient access to good nutrition or food to eat their fill. People
would have to make more frequent visits to grocery stores to acquire their food,
thereby increasing their chance of infection. Disadvantaged populations living in
food deserts may face a greater risk of malnutrition retailers hoarding nutritious
foods because of the COVID-19 epidemic. Substandard nutrition may make this
more of an issue in food-deprived areas than in wealthier areas.

4.6.3.2 Local/global surrounding conditions

How can you handle the topic of “air quality, water quality, infrastructure, and
pollution?” Health inequalities may be quantified by looking at when people wind
up in specific neighborhoods and where they are located. Non-whites are most
likely to live along the ocean coastlines [27]. In high-poverty neighborhoods where
minorities with low economics are concentrated, pollution seems to be caused by
industrial polluters like refineries, landfills. Groundwater was determined to be the
principal source of drinking water for about a third of the population. Foundries,
refineries, and garbage dumps usually contain hazardous waste. It has been pro-
posed that pollution in the environment may render people more susceptible to
COVID-19. It is thought that pollution particles are a mode of transportation for the
virus. The new study claims show that air quality might have increased the infec-
tion rates. This may be that air contamination makes one’s immunity weaker,
allowing them to contract infections. As recently determined, a fine-particle level
rises, known as PM 2.5, will facilitate the spread of COVID-19 A analysis showed
that there was a correlation between a rise of 1 mm/m3 and 8% of deaths due to
COVID-19 [52]. Health still has great importance. Residents of high-poverty
communities are less healthy because of their poverty. Many people will stroll,
bike, or work out than play in the open area nowadays. The next problem in high-
poverty communities is that residents in these neighborhoods do not have access to
greenery. These neighborhoods are so well populated that it is hard for their inha-
bitants to find room to be green. Increased social distance also helped one to decrease
the COVID-19 infection rate. In neighborhoods where there are many people, dis-
tancing yourself from the crowd is realistic. You are more likely to get the infection if
 Impact of world pandemic “COVID-19” 75

you live in densely populated areas and spread it. Poorly maintained low-income
households also inhabit public housing. Too many infestation problems of cock-
roaches, rodents, rats, and other vermin are conducted and discovered. Some things
often seen in these conditions were mold, the absence of air conditioning, smoking,
and cigarette use. Children who grow up in a housing project are twice as likely to be
born with asthma than children who grow up in a single-family household. People
living in poverty could be susceptible to COVID-19.

4.7 Academy

4.7.1 High school graduation

For most occupations and college and further education, a high school diploma is
mandatory [53]. With no high school diploma and no educational opportunities,
work opportunities are restricted. If there are not enough work openings, people
will fall into poverty. Poverty has also been debated to result in unfavorable health
effects. It is often the home and school that determines if a high school student
graduates. Studies show that children who have limited contact with their parents
when in high school are far more likely to drop out. Schools that have a higher
degree of criminality are more likely to have lower attendance and lower educa-
tional achievement. Students from lower income families are less likely to have
money and thus therefore are less likely to get good schooling. At the time of the
COVID-19 epidemic, many schools had to go online to avoid the disruption of
face-to-to-face schooling. These children do not have Internet connectivity. The
theory behind this finding is that children from upper class households have more
time flexibility to access and use online resources than their lower class counter-
parts. Studies more extended, but some can close the door to opportunities early on.
An uneducated parented child can value education below that of an educated one
and downplay academic requirements, thereby giving it a disproportionate low
value. This should not affect the value of parents with a higher degree of education.
Education is necessary regardless of any difficulties or barriers that the individual
can face.
Non-educated parents are letting their children miss school to do their home-
work while the others are spending hours online. Having to engage in distance
learning will affect academic success for specific children.

4.7.2 Psycholinguistics and literacy

Minorities and individuals with more inadequate degrees of schooling are most likely
to be included in working-class neighborhoods. Studies have shown that people with
language and reading barriers have more severe health problems, do not have
insurance, and have a more challenging time following orders on a drug regimen.
The United States is host to numerous non-English speakers. Medical students and
physicians at Harvard also launched a new project dubbed the “COVID-19 Health
Literacy Project.” This knowledge that is part of COVID is now known in
76 The Internet of medical things

35 different languages [54]. The capacity to translate into languages includes Arabic,
Bengali, Chinese, Dutch, Hindi, and several others. The virus details, ways to avoid
catching the virus, and any remedies for illnesses are available on the fact sheets.
This has created an open forum for discussion also for non-native speakers. It is vital
to increase public knowledge of the infection and promote safety measures to stop
the spread of the disease.

4.8 Stable, predictable employment

Money, the work one has, and incentives such as healthcare, compensated medical
time, and family leaves play a significant role in this equation. Workplace racial
inequalities often occur. Then Caucasians to occupy positions, African-Americans,
and minority groups are overrepresented in the blue-collar workforce. Psychological
discrimination in the workplace may have devastating results, leading to fatigue,
anxiety, and poor health. Unemployed individuals are at greater risk for stress dis-
orders such as hypertension, diabetes, and CVD, making them more susceptible to
COVID-19. The US economic performance has decreased after export and import
restrictions were announced. For the last 7 years, an estimated 33.5 million
Americans have applied for jobless benefits [55]. When people are out of work, they
risk or forfeit their benefits. Allowing someone who cannot afford care to be
screened for the COVID-19 virus has been ineffective. On March 27, 2020, the
President signed into law of the Coronavirus, a Food, Relief, and Economic Security
Act, also known as the Hope Act [56]. It gives $1,200 to any American family with
an income of $75,000 or less for a stimulus [57]. The family gets $1,200, and there
are two children under 18 years of age. That increases to $1,500 for each minor boy.
However, to obtain a stimulation check, the conditions may appear clear on the
surface, which are various and complex. It has not been determined how the allo-
cation of funds would be carried out. As of this moment, the individuals who apply
for the stimulus search have been sent one. Speaking to both the President and
Congress about issuing a second search has been proposed, but no one knows when
to follow through. Many families will not get away with a single $1,200 check. This
will probably be a hardship for households, as when the crisis is over and everybody
will go back to work until they get paid.
The economic downturn and the distribution of COVID-19 (on one side); you
must tread lightly to decrease COVID-19. A combination of social isolation and
crowding social distancing outward seems to be suitable for making the curve more
stable. Still, it comes at the cost of the country’s financial prosperity. Each state’s
governor proposed civic distancing, and so a period of social segregation ensued.
The rise in March and May month events and a drop in June and increase in July is
an excellent example of regression toward the mean. Georgia Governor, Gillespie
declared a state of emergency on March 25, though the Idaho Governor did not
declare one until 1 month later, on April 3, 2020 [58]. Staff who were not necessary
during the lockdown were ordered to remain home and just go to the food store or
drugstore if they are necessary. The whole station was placed on lockdown before
 Impact of world pandemic “COVID-19” 77

social distancing had been executed. The country’s emergency communications

system was made operational on April 24, 2020 [59]. During April, the cases in
Georgia began to rise slightly. Preferably until the lockdown was applied, by June
of 2015, the occurrence in Georgia was higher. On the other hand, the Governor of
Idaho declared a State of Emergency on March 25, 2020 [58]. Over the first few
weeks of the month, it increased, and by the middle of the month, it had subsided.
The state was finally unlocked on April 30, 2020 [60]. In mid-April to the start of
June, there were <40 recorded cases of Lyme disease in Idaho. After the date June
1, 2020, when it was first put in place, the frequency is increasing, and the increase
is more significant in June than in July. It is seen in Figure 4.2 that the average
prevalence of COVID-19 cases in the United States is substantial. It is clear that the
new trend of frisking, specifically, has been reestablished as the number of
instances around the country has returned to its pre-lockdown rate as it was before
the locks being installed and has continued to spread out in that manner
(Figure 4.2). It is reasonable that enforcing a lockout around the nation’s borders
takes priority given the national economy. In the meantime, we are working to

60,000
March April May June
10,000

20,000

0
USA
15,000
March April May June
Number of new COVID-19 cases/day

10,000

5,000

0
New york
4,000
March April May June
3,000
2,000
1,000
0
Georgia
400
March April May June
300
200
100
0
Idaho

Figure 4.2 The wax and wane in new cases of COVID-19 per day in USA, New
York, Georgia and Idaho. The graphs were generated using the online
data form CDC and John Hopkins web sites
78 The Internet of medical things

eliminate COVID-19; the national economy would have to be shown what it has
in store.

4.9 The impact of the COVID-19 pandemic on firms

The new COVID-19 is a growing threat to the global economy, public health, and
national security. As of June 30, 2020, there have been ten million reported cases,
causing about 500,000 deaths in 215 countries (as per reports from the WHO and
Johns Hopkins) [61,62]. I predict that the COVID-19 crisis would trigger sig-
nificant problems in both infectious disease and mass public health outbreaks and
extensive food poisoning in the community. According to an Organization for
Economic and Co-operation Development (OECD) estimate, the global GDP
growth rate is projected to decline to 2.4% by 2020. The possibility of instability,
and a significant decrease in global economic activity due to sharp drops in
production and spending and supply chain disturbances has risen. This is a huge
challenge for many countries as well as for governance and cooperation around
the world. Many people being sick could have a major effect on China’s econ-
omy [63–65]. As of April 17, China’s gross domestic product had risen 13.5%,
investment, and usage by 24.5%, and all dropped by 15% compared to the year
before; however, Chinese people’s unemployment increased to 6.2% for the first
time in the first quarter, according to the figures published by the National
Bureau of Statistics of China.
Entrepreneurship is fundamental to the country’s economy as a whole.
Because of this, for companies, it is imperative to consider the current state of firms
and their coping mechanisms and how the effects of the COVID-19 pandemic on
them are. As soon as the disease epidemic has run its course, researchers will begin
to study corporations’ innovation plans and responses [66–68]. Other sectors have
also been studied to identify possible solutions to the pandemic. However, these
findings have not been supported by observational evidence.
Firms in Guangdong Province were examined to see how they were affected by
the COVID-19 pandemic and if any policy responses could be prepared (the pro-
vince with the highest GDP in China). To explore why the pandemic impacted
corporations, they went on to look into whether the disease has affected them, and
what they expected in the future. Any supportive measures can be implemented.

4.9.1 Time-saving method

This study was reviewed by the school of management’s IRB, with approval
granted in the form of IR2020. It was decided that a survey will be done in
Guangdong Province. According to the following formula, the minimum sample
size was determined. From the previous information, the total minimum population
was estimated to be 384. Nonetheless, the survey was expanded to 500 to boost the
overall accuracy of the data set. Distributing in Dongguan, Shenzhen, along with
many city government offices, there are 21 cities in Guangdong Province and the
GDP of the 15 cities included in our sample represent more than 90% of the total
 Impact of world pandemic “COVID-19” 79

GDP of Guangdong Province. Firms on this list were chosen according to

Guangdong province’s industry characteristics and the scale of the industry.

4.9.2 Data gathering

The study was done between April 10, 2020 and April 25, 2020 while everybody
was out on vacation. Three company executives were reached via telephone or
WeChat, a mobile app. The survey was entirely voluntary and we received
informed permission from all participants. In the survey, the important research
priorities, participant confidentiality, and ethical principles were outlined to parti-
cipants well in advance of the start of the survey. As much of the queries as they
might, they were required to. Even, they were not confident and could avoid
making an effort to choose a response by leaving it blank. Altogether, 1,553
questionnaires were given.

4.9.3 Computations and statistical review

Entering the data was needed. No more data analysis was needed. People were
eliminated from the survey because they responded with less than 30% of the
questions correctly and lacked relevant information. Descriptive questions were
used to get information about the sample’s demographic profiles and describe the
findings.

4.9.4 Statistical results

Characteristics of the demographic group: An answer rate of the survey was found to
be a success at 94.8% (524 of the population replied). The several sectors in the
business analysis included information technology (18%), engineering (34.6%),
banking (12%), service and commercial (7.4.4%), and industrial (12.6%), business
and construction (14.2.6%), and trade (7.4%), 4.8% of the GDP have been affected.
This manufacturing distribution follows Guangdong’s economic developments clo-
sely. As polled, the businesses were comparably sized. A total of 141 businesses, out
of the total of 300 and 6,600, made up around 26% of the overall firms and had 50 or
fewer workers. In contrast, 162 (the half of the 500-employee category, and over) out
of the total of 5,000, the total of 500, accounted for around 24% of the total number
of companies.
Firms would be great because of the effect of the pandemic. Since more than
half of the companies carried out their day-to-day activities but suffered stock-
running problems, some were shut down or remained operational due to shortfalls
of materials. Many companies face at least one of these challenges: wage and social
security, rent payment, customers’ overdue payments. Though the cost of labor and
cost of production tendering are important for larger companies, firms with less
than 50 workers rent appear to be the main driver. The vast majority of companies
sustain output barely, with materials or materials’ scarcity. And 22.9% of compa-
nies received domestic requests, although 63.9% experienced delays or unable to
ship their product on schedule.
80 The Internet of medical things

4.9.4.1 Gibberish
Redeems to concentrate on demand patterns in the public sector and behavior
improvement in the public sector. Pandemics would lead to less GDP and jobs, con-
sonant with our research on the COVID-19 pandemic. This research, however, just
looked at aggregate trends without addressing the companies, which better captures
the effect of the COVID-19 pandemic on the economy. Firms would be significant
because of the effect of the pandemic. Returning to full manufacturing capability was
difficult because of the disease. More than half of the businesses reported problems
with inventory; some of them also claimed they were completely out of material. One-
third of the businesses we surveyed are feeling operational stresses, so we would
estimate that they hire one-third of the workers and incur one-third of the costs. Also,
23.4% of businesses have done away with domestic requests.
Many businesses stated that market penetration was impeded because of client
face-to-face and location meetings being challenging to conduct. Furthermore,
nearly two-thirds of the companies posted increased operational costs, including
postponed orders, distribution taking more prolonged, and less-than-expected pro-
duction output. Much of them had to contend with the financial problems asso-
ciated with their separation. They were still being held on strict accounts for payroll
salaries, insurances, leases, financial obligations, and accounts. As shown by the
answers to question 4 (supply of raw materials, spare parts, other production and
processing materials), the business owners explained that the supply chain is
affected by epidemic-related delays at varying degrees across the regions. Due to
the spread of the disease, several export industries were threatened. Several busi-
nesses announced cancellations and postponed delivery of overseas orders.

4.9.4.2 Companies dealt with the outbreak

A new pandemic and the rivalry with the United States have sparked further R&D
investment and creativity in the Chinese economy. In the words of Creative and
Venn, more than half of the companies agree that this global pandemic has helped
the growth of virtual workplaces, remote workers, and market interruption methods
and strengthened their capacity to deal with major disasters. A few people sug-
gested that an outbreak would take out rivals in the long run, leading to company
launches or expediting improvements in marketing strategies (such as community
channel expansion). As an example, the company’s primary market retailers
include larger shopping centers. They relocated staff from their stores to online
sales, which enabled the company to focus on online markets throughout the crisis.
When they did this transition, they had outstanding crisis in hand thanks to utilizing
social networking and live-streaming channels. It has done the trick, as certain
companies have grown as a result of the outbreak and others that have adopted it to
try new industries. Around the same time, it has helped customers become more
conscious of their well-being and led to behavioral changes in consumption. Firms
also discovered that an increase in consumer interest in creativity has created new
business possibilities and fueled their changes in marketing practices. Several
interview panel members suggested an imperative, for faster growth, for example,
 Impact of world pandemic “COVID-19” 81

online schooling and 5G. Any companies also replaced traditional store-based
methods with multichannel marketing. As an example, on the other hand, the shop
is closed, but community shopping and social marketing are taking place.

4.10 Healthcare, social and economic challenges in

Bangladesh

Reported cases and fatalities are projected to be 48,786 and 1,482 on November 7, 2020.
Coronaviruses are the first known viruses that were detected in Wuhan, China. These
viruses mutate rapidly and have a great mutation risk of infection. The main problem
caused by coronavirus infection is severely acute respiratory illness. The number of
people infected by this lethal virus has increased from around 219, increasing the inci-
dents. The WHO announced that there is a pandemic of COVID-19 because it has spread
all over the globe. North America and Europe have both been identified as having nearly
21,000 cases of COVID-19, while Europe has 12,000 registered in secondary institutions.
The four nations with the most COVID cases on Earth up to November 7, 2020 are USA
(9,504,758), India (8,462,080), Russia (1,733,025), and France (encompassing the same
number of cases), and Russia (9,503,758). About 45,000 people have died in China due
to the disease and over 91,000 have been diagnosed with it [69]. And the extension of
deadly diseases from another lower middle class Asian nation with a population of
161.3 million [70], Bangladesh is caught in the tragedy. Bangladesh was the first to
recognize Creative Origins on March 8, 2020. Until the end of the date, the total of
COVID-19 deaths has been set at 410,988 and the number increases to 5,966.

4.10.1 Methodology
This chapter also included qualitative and quantitative analysis methods in some of
the research approaches used in it. Data from secondary sources, writers’ forecasts,
details from the WHO, and related academic institutions (IEDCR) were used.
Official data estimates have been used to calculate the population’s vulnerability.
Thus, other polls and tests have been used to confirm these results. Such personal
impressions of the world around the United States will be supplemented with data
collected from a wide range of different outlets and anecdotal knowledge from all
over the country. Primary research done in PubMed and Google Scholar was done
on COVID-19. Healthcare, religious, and economic data on Bangladesh was
gathered from March to November 7, 2020. Keywords were used in the quest, e.g.
“COVID-19,” “Coronavirus outbreak.”

4.10.2 Analysis of basic healthcare

4.10.2.1 Healthcare service
Following the WHO guidelines, Bangladesh has implemented policies to reduce
the transmission of COVID-19 to minimize the transmission. Among the policies,
included are mandatory use of facial masks, limitation of travel, social distancing,
and lifestyle modifications. In addition, the government and private sectors are
82 The Internet of medical things

striving to promote good personal hygiene by utilizing local media. A research

published by Farhana and Mannan [71] demonstrated that a substantial number of
Bangladeshis were unaware of, not aware of, and entirely confused about the pro-
pagation, indications, and incubation time for COVID-19. Bangladesh hospitals–
treated patients faced a great deal of difficulty as a result of the emergence of
COVID-19. Serious disadvantages were discovered in facilities providing ventilation
for cases of acute severe respiratory illness. For every 1,000 residents, there are
around 8 ICU beds. According to the WHO condition study, as of October 26, 11,730
available beds and 564 ICUs are ready to handle COVID-19 patients nationwide. In
Dhaka, 35.19% of all hospital beds are available to the general population and the
remaining beds and ICUs are dedicated to medical patients. Due to many infected
patients being treated at home, beds in the hospital were not filled [72].
At first, the facilities were not sufficient but with time, they grew. Over time,
more and more government and private healthcare providers have come to use C.O.
V.I.D.T.s (or C.O.V.D.D.). It must be emphasized, though, that the COIMEST’s
research scope in Bangladesh, at only 14.6 million tests, is considerably less than
that of the countries in South Asia. Over most of the 1980s and 1990s, WHO
sponsored MOHFW to increase testing capability and currently are underway plans
for further development in the nation. In this latter case, WHO also launched a
comprehensive effort to ship out hundreds of samples from all of the 64 districts of
Bangladesh. With WHO’s help, as of September 2020, nearly half of all the sam-
ples were distributed to laboratories throughout the region [73].
Following a coronavirus outbreak in Bangladesh, the GoB charged a fee of 200
taka (80 pence in United Kingdom currency) for the COVID-19 test instead of the
3,500 taka charged by the private sector. Also, proportionally, the occurrence of
research has decreased to about 0.8 tests per 1,000 individuals each day. The doctors
at the CDC have made personal commitments to assist and heal people even when
they have the moral obligation to look after their safety and security. The medical
staff and other frontline providers of healthcare have difficulty treating the COT-
treated patients appropriately because of a lack of sufficient equipment, a limited
capacity for supporting them, and the Government’s lack of readiness to handle the
crisis. This coronavirus has caused the deaths of countless healthcare workers
worldwide, especially in Wuhan, China. Physicians and healthcare staff are particu-
larly susceptible to this deadly disease in Bangladesh [74].

4.10.3 Precarious living

Due to a lack of resources, the cessation of municipal facilities and everyday
operations culminated in a medical emergency. Around one-fifth of the nation lives
in poverty and a huge percentage of the workforce is engaged in part-time
employment that does not pay. Loss of work vs loss of lives, the shutdown pre-
sented a quandary. It is estimated that about 62 million people in Bangladesh are on
wage or salaried jobs, with 10 million in casual employment made up of those
working on regular wage/salary. At least 4.5 million regular laborers work in the
construction, transportation, and lodging industries who were deeply impacted [75].
 Impact of world pandemic “COVID-19” 83

Around 4 million citizens in the world depend on the RMG industry. As

developing countries, which were big importers of this industry, began cancelling
orders for COVID; the livelihoods of those employees were threatened and the
remaining 4 million people work in the manufacturing industries, of which 85% are
working informally. Employees are at the same rate of pay as others who do not
come on as they have no special skills. Additionally, many self-employed jobs in
the informal economy are often self-employed individuals, numbering over
5.19 million people. A total of 18 million people’s life was under threat throughout
the shutdown era. It will seem that 72 million families in America are currently
suffering economic hardship. The package it adopted for the promotion of the
economy consists of lending to export-oriented businesses for a total of Tk 117
crore for salaries, as well as making various varieties of loans to small and medium
companies and groups who have sent their employees overseas, to unemployed
people and staff who have returned from abroad, and job postings on both of the
country’s cash and kind (welfare support) to the poverty stricken. However, there
have been considerable execution problems around the different schemes, with big
companies performing better than local, grassroots, family, and micro-businesses.
The cash benefit was modest, and there were flaws in the administration. This has
not been shocking, according to numerous studies, that the rate of poverty rose
during the crisis [76,77].

4.10.4 Social distancing

Lack of involvement in social interaction has become the principal approach to
monitor and may even reverse the influence of COVID-19 infection [78].
There is no known means of treating morbidity and fatality, so humanity must
preserve social distance. Regulation of social-distancing requirements becomes a
significant concern for large numbers of the population in Bangladesh who rely on
their regular employment. About 55% of the population in urban Bangladesh is
believed to live in slums. Around one in five people in Third World families live
in a slum, according to the OECD. Slightly over half of the respondents in
Chittagong’s slums have four to six people in a room while the other quarter has
seven to ten people to share a room (2018). It is typically difficult to access
individual rooms in the shantytowns since the streets are typically too small for
two people to walk side by side. Compliance can also be challenging for others
who are eager to go along. A regular laborer, a dealer, and rickshaw-pullers having
little resources must distance themselves from their communities [79]. Some
institutions also questioned the significance of social exclusion in the broader
social context as well. If resources are distributed equally that completely depend
upon which would depend on who is in-line to receive them, who can afford to
wait in-line, and what is distributed free of charge by social aid. Socioeconomic
considerations can affect conformity with social distance norms. Since too many
drivers are using the same mode of transport, the traffic gets much worse at
holiday and festival time. Public understanding of social distancing could have
been hindered by missing social data [80].
84 The Internet of medical things

4.10.5 Groups with special needs

People turn to religion when stressed, but you must make them see it unemo-
tionally as a way of keeping distancing yourself from it. People would instead go
to churches to seek hope and a sense of community. We found two significant
violations in adhering to the Lockout Protocol. The first was an excellent
opportunity for people from Dhaka to come home because it saved them from a
career in the RMG industry. The second was to accompany a religious leader’s
funeral. However, various constituencies have assailed the GOOP with accusa-
tions of managerial incompetence [80–82]. Each network team that implemented
this kind of application customized it for their corporate purposes, with various
inputs from different stakeholders. Still, overall, few groups do not benefit
from incorporating one: (differentiation of channels for) flexibility in their
application software.
A higher degree in deterioration of both physical and mental well-being is
associated with the shutdown. The spread of coronavirus multiplies the demand for
mental and physical well-being through elderly, disabled, economically dis-
advantaged ones, and the sick ones are more severely impaired. Domestic abuse
and gender-based harassment is at the forefront of media attention due to a world-
wide surge in the use of restrictions, as reported in both local reports and on social
networks. More and more companies also moved work away from the office to the
workplace, with women having to perform their daily domestic duties longer hours
as well. It has been established from various sources that the level of aggression has
been rising in the home is becoming disturbing.

4.10.6 Discussion
To address the widespread proliferation of COVID-19, GoB has implemented
several prevention and control measures. In Bangladesh, MOHFW set out
“Bangladesh Emergency Health Preparedness and Response Strategy for COVID-
19.” The foremost concern of this strategy is to slow the progression of the illness.
To further improve the ICU use, the hospital built a new facility, hired, and
equipped 2,000 more physicians and employed 5,000 nurses. The GoB initiative
guaranteed surgical devices, supplies, and health-related services for COVID-19
patients and providers. However, they do not have a higher number of patients
relative to their counterparts in other [nearby] countries No well-equipped COVID-
19 research laboratories and no clinical or health services for these citizens in the
country of Bangladesh make finding a cure difficult. This, thus, shall ensure that
the GoB has proper research facilities and health treatment for this lethal virus. For
many impoverished citizens in Bangladesh, it was a life-or-or-death decision to
take the long-malaria test. As opposed to staying sealed, more people choose to
consider the latter to ensure their long-term survival. The shutdown caused millions
of Americans to lose their jobs and wages.
It was hard enough and arrived too late. As a result, a lot of people who were
not well-off have been much worse off. Social distancing is difficult because of the
socioeconomic, environmental, and economic challenges that citizens face in
 Impact of world pandemic “COVID-19” 85

Bangladesh. Furthermore, lacking understanding, negative feelings, and little con-

cern for social distancing has been putting the country at risk for the spread of the
COVID-19 virus. Poverty and the fear of unemployment are legitimate reasons for
noncompliance. Due to the nature of the chemical agent, wearing a respirator is
critical in containing exposure to COVID-19 and therefore must be done at all
times. According to a survey, it is estimated that approximately 63% of the popu-
lation of Bangladesh wears a mask. About 53% of Dhaka division residents use
facial masks to prevent the spread of infectious diseases. The study confirms that
residents in metropolitan environments do not adhere to the health recommenda-
tions to avoid transmitting disease. So for the public to be informed of this virus
presents a problem to the government. During potential pandemic times, inter-
ventions might include social isolation and wearing masks in public. In periods
with increased parental supervision, domestic abuse occurs more often. Not just
poor people, but mothers at all income levels face self-esteem problems as they are
forced to stay home full time to support their families. Bangladesh is notorious for
its hatred and brutality against women and children. About 35% of married women
experience sexual abuse at the hands of their spouses. Despite laws and recognition,
women continue to encounter a substantial obstacle in their pursuit of fair repre-
sentation in public life. The factors that need to be considered when formulating a
plan for healthcare communication strategy are communication (meaning which
will be communicating with patients and how), appropriateness (how people can
perceive and relate their message to the treatment provided), value (a treatment can
mean a lot of things to a patient), and significance (which it is most significant to
the patient).

4.11 Conclusions
At the COVID-19 pandemic’s advancing around the globe, the COVID-19 death toll
is accumulating. Aside from the increasing amount of cases and fatalities, also on the
whole, the epidemic had a dangerous side impact on economies: In addition to
decreasing their numbers, it increased their fragility. Following the announcement of
pandemic status in March 2020, bans on global travel were put in place, with citizens
encouraging anyone who could avoid being infected to avoid public areas by work-
place exposure and ensuring that they were at home at all times. In late March of
2006, Wuhan, China, was where the first cases of the disease were discovered. In
Brazil and the LMICs, a significant number of people in the United States, but also
the United States, India, and other higher income countries, were hit by the epidemic.
Unpreparedness was a big part of the problem of healthcare services. In some
instances, healthcare staff experienced a health-productivity gap. COVID-19 failures
also spurred healthcare institutions to find new necessities for patients. Telemedicine,
psychological distancing, masking, and handwashing have also been seen to lower
the burden of the COVID-19 pandemic.
Pandemic influence has more to do with other people than health services.
The community that resides in poverty and is heavily polluted with sewage and
86 The Internet of medical things

garbage is being hit harder by this toxin. Additional assistance, such as a benefit
payment for the poor, is needed. During epidemic outbreaks, it is highly critical
for this group to have a careful eye on this demographic since they are more
susceptible to disease. You become more susceptible during epidemics because of
SES and ethnic classification. Dietary patterns and low wages are correlated with
lower SES. Latinos are more disadvantaged than Caucasian populations owing to
personal and structural inequality, and as a result, they are more prone to suffer
from ill-health effects. Therefore, it is clear that the pandemic has passed them
by, perhaps never yet suspected. Richard Clarke Cabot, an American physician,
thought of both of these aspects while doing his daily work. On the basis of these
observations, he concluded that people with worse health were more likely to die.
This research supports the conclusion that poverty, disparities, and SDOH
increase the transmission rate of illness. While inequalities in health and disease
cause disparities in morbidity and mortality, inequities in healthcare exacerbate
the problem. Current research on epidemics has paid no attention to the impacts
of social inequality on well-being at periods of an epidemic, which means they
missed the point entirely, because they failed to do both. Because of this, it is
essential to respond to any health emergency promptly. To avoid an epidemic of
illness, it is essential to be aware of all the well-being and healthcare causes.
Identifying specific health and healthcare causes, such as SDOH, will aid in
providing access to a comprehensive set of services to the socioeconomically
vulnerable. Education of the public on the virus seriousness of the disease and
awareness-inducing nature of the illness is equally necessary. If you are more
aware of the community that is at risk than others, it may affect the decision to
avoid others that are less. Taking into consideration, the many causes will help
one to stop an epidemic from increasing. Appropriate and timely clinical treat-
ments, together with well-directed and timely schooling, have proved to be suc-
cessful countermeasures to the COVID virus. Reducing disparity between well-
being and healthcare may be achieved by using SDOH. Such projects need an
interdisciplinary approach that comprises doctors, anthropologists, public health
officers, and researchers, along with others from the National Institutes of Health,
the WHO, and others, to thoroughly understand various causes of health dis-
parities that groups of people face. We will have to determine how the last one
goes because we can deal with potential plagues.
Based on COVID-19 pandemic knowledge, a variety of hypotheses may be
reached. Asking the correct questions is critical in dealing with such a problematic
issue. Thus, rather than predicting the task’s difficulty, it is more prudent to assume
that it is essential. Second, more stringent policies and efficient compliance are
necessary if an epidemic is to deter or slow down its dissemination. Third, taking
note of the shortcomings of the health sector revealed by this experience, effective
steps must be taken to enhance public health. Proper plans, including improved
equipment of health facilities and staff training, are crucial in this respect. To avoid
the further transmission of COVID-19 infections, stringent controls on cross-
infections in hospitals should be taken.
 Impact of world pandemic “COVID-19” 87

References
[1] He X, Lau EHY, Wu P, et al. Temporal dynamics in viral shedding and
transmissibility of COVID-19. Nat Med. 2020; 26: 672–5. https://doi.org/
10.1038/s41591-020-0869-5.
[2] Li Q, Guan X, Wu P, et al. Early transmission dynamics in Wuhan, China, of
novel coronavirus-infected pneumonia. N Engl J Med. 2020; 382: 1199–207.
https://doi.org/10.1056/NEJMoa2001316.
[3] Heron M. Deaths: leading causes for 2017. National Vital Statistics Reports,
2019; 68(6).
[4] Congressional Budget Office. CBO’s current projections of output,
employment, and interest rates and a preliminary look at federal deficits for
2020 and 2021. Congressional Budget Office (2020). https://www.cbo.gov/
publication/56335 (accessed October 25, 2020).
[5] Wu Z and McGoogan JM. Characteristics of and important lessons from the
coronavirus disease 2019 (COVID-19) outbreak in China: summary of a
report of 72314 cases from the Chinese center for disease control and pre-
vention. JAMA – J Am Med Assoc. 2020; 323: 1239–42. https://doi.org/
10.1001/jama.2020.2648.
[6] National Collaborating Centre for Determinants of Health. Marginalized
populations. National Collaborating Centre for Determinants of Health
(2020).
[7] Yang Song, Min Zhang, Ling Yin, et al. COVID-19 treatment: close to a cure?
A rapid review of pharmacotherapies for the novel coronavirus (SARS-CoV-2),
International Journal of Antimicrobial Agents, 2020; 56(2): 1–8.
[8] Bhuyan HK, Chakraborty C, and Shelke Y, Pani SK. COVID-19 diagnosis
system by deep learning approaches, Expert Systems, July 2021 (Early
Published).
[9] Czeisler M-E, Lane RI, Petrosky E, et al. Mental health, substance use, and
suicidal ideation during the COVID-19 pandemic—United States, June
24–30, 2020. MMWR Morb Mortal Wkly Rep. 2020; 69: 1049–57. https://
doi.org/10.15585/mmwr.mm6932a1.
[10] McGowan ML, Norris AH, and Bessett D. Care churn – why keeping clinic
doors open isn’t enough to ensure access to abortion. N Engl J Med. 2020;
383: 508–10. https://doi.org/10.1056/NEJMp2013466.
[11] Satiani B and Davis CA. Practice management the financial and employment
effects of coronavirus disease 2019 on physicians in the United States
(2020). doi:10.1016/j.jvs.2020.08.031.
[12] McMichael TM, Currie DW, Clark S, et al. Epidemiology of COVID-19 in a
long-term care facility in King County, Washington. N Engl J Med. 2020;
382: 2008–11. https://doi.org/10.1056/NEJMoa2005412.
[13] Sullivan-Marx E. Aging in America: how COVID-19 will change care,
coverage, and compassion. Nurs Outlook. 2020; 68: 533–5. https://doi.org/
10.1016/j.outlook.2020.08.013.
88 The Internet of medical things

[14] Normile D, Cohen J, Enserink M, et al. As normalcy returns, can China keep
COVID-19 at bay? Infected travelers pose a continuing threat, but local
coronavirus transmission still occurs as well. Science. 2020; 368: 18–9.
https://doi.org/10.1126/science.368.6486.18.
[15] Varghese GM and John R. COVID-19 in India: moving from containment to
mitigation. Indian J Med Res. 2020; 151: 136–9. https://doi.org/10.4103/
ijmr.IJMR_860_20.
[16] Bhuyan HK and Ravi VK. Analysis of sub-feature for classification in data
mining, IEEE Transaction on Engineering Management, 2021 (Early
Published).
[17] Statista. COVID-19 deaths per capita by Country. Statista (2020).
[18] Lee VJ, Chiew CJ, and Khong WX. Interrupting transmission of COVID-19:
lessons from containment efforts in Singapore. J Travel Med. 2020; 27: 1–5.
https://doi.org/10.1093/jtm/taaa039.
[19] Singapore’s economic contraction in the second quarter was worse than
initial estimates, August 2020, https://www.cnbc.com/2020/08/11/singapore-
releases-second-quarter-2020-gdp-economic-data.html.
[20] Bhuyan HK, Chakraborty C, Pani SK, and Ravi VK. Feature and sub-feature
selection for classification using correlation coefficient and fuzzy model.
IEEE Transaction on Engineering Management, May, 2021 (Early
Published).
[21] Daniel G, Stefania F, and Johannes W. COVID-19: Without Help, Low-
Income Developing Countries Risk a Lost Decade, August 27, 2020. https://
blogs.imf.org/2020/08/27/covid-19-without-help-low-income-developing-
countries-risk-a-lost-decade/
[22] Suleiman A, Bsisu I, Guzu H, et al. Preparedness of frontline doctors in
Jordan healthcare facilities to COVID-19 outbreak. Int J Environ Res Public
Health. 2020; 17. https://doi.org/10.3390/ijerph17093181.
[23] Ahmed SAKS, Ajisola M, Azeem K, et al. Impact of the societal response
to COVID-19 on access to healthcare for non-COVID-19 health issues in
slum communities of Bangladesh, Kenya, Nigeria and Pakistan: results
of pre-COVID and COVID-19 lockdown stakeholder engagements.
BMJ Glob Health. 2020; 5: e003042. https://doi.org/10.1136/bmjgh-2020-
003042.
[24] Martinez D, Sarria GJ, Wakefield D, et al. COVID’s impact on radiation
oncology: a Latin American survey study. Int J Radiat Oncol Biol Phys.
2020; 108: 374–8. https://doi.org/10.1016/j.ijrobp.2020.06.058.
[25] Luedi MM, Doll D, Boggs SD, et al. Successful personalities in anesthe-
siology and acute care medicine: are we selecting, training, and supporting
the best? Anesth Analg. 2017; 124: 359–61. https://doi.org/10.1213/
ANE.0000000000001714.
[26] Li Q, Guan X, Wu P, et al. Early transmission dynamics in Wuhan, China, of
novel coronavirus-infected pneumonia. N Engl J Med. 2020; 382: 1199–207.
doi:10.1056/NEJMoa2001316.
 Impact of world pandemic “COVID-19” 89

[27] Rothan HA and Byrareddy SN. The epidemiology and pathogenesis of cor-
onavirus disease (COVID-19) outbreak. J Autoimmun. 2020; 109: 102433.
doi:10.1016/j.jaut.2020.102433.
[28] CDC. Preliminary estimates of the prevalence of selected underlying health
conditions among patients with coronavirus disease 2019—United States,
February 12–March 28, 2020. MMWR Morb Mortal Wkly Rep. 2020; 69:
382–6. doi:10.15585/mmwr.mm6913e2.
[29] AMA. Social Determinants of Health, Health Systems Science Learning
Series. Chicago, IL: AMA (2020).
[30] CDC. Coronavirus Disease 2019 (COVID-19): People Who Are at Higher
Risk for Severe Illness. Atlanta, GA: CDC (2020).
[31] Chapman AR. The social determinants of health, health equity, and human
rights. Health Hum Rights. 2010; 12: 17–30.
[32] Watson K. Why schools in rich areas get more funding than poor areas.
Global Citizen (2016). Available online at: https://www.globalcitizen.org/
en/content/cost-of-education-in-us/
[33] ODPHP. Social cohesion. In: Healthy People 2020. Washington, DC: U.S.
Department of Health and Human Services (2020).
[34] Kluch Avas. Americans’ social distancing beliefs and activity (May, 2020).
Available online at: https://news.gallup.com/opinion/gallup/311261/amer-
icans-social-distancing-beliefs-activity.aspx.
[35] ODPHP. Discrimination. In: Healthy People 2020. Washington, DC: U.S.
Department of Health and Human Services (2020).
[36] Kelli HM, Hammadah M, Ahmed H, et al. Association between living in
food deserts and cardiovascular risk. Circ Cardiovasc Qual Outcomes. 2017;
10: e003532. doi:10.1161/CIRCOUTCOMES.116.003532.
[37] Mackenbach JP, Cavelaars AE, Kunst AE, and Groenhof F. Socioeconomic
inequalities in cardiovascular disease mortality; an international study. Eur
Heart J. 2000; 21: 1141–51. doi:10.1053/euhj.1999.1990.
[38] Banerjee T, Crews DC, Wesson DE, et al. Food insecurity, CKD, and sub-
sequent ESRD in US adults. Am J Kidney Dis. 2017; 70: 38–47. doi:10.1053/
j.ajkd.2016.10.035.
[39] Osuna-Padilla IA, Leal-Escobar G, Garza-Garcia CA, and Rodriguez-
Castellanos FE. Dietary acid load: mechanisms and evidence of its health
repercussions. Nefrologia. 2019; 39: 343–54. doi:10.1016/j.nefroe.2019.
08.001.
[40] Abbasalizad Farhangi M, Nikniaz L, and Nikniaz Z. Higher dietary acid load
potentially increases serum triglyceride and obesity prevalence in adults: an
updated systematic review and meta-analysis. PLoS One. 2019; 14:
e0216547. doi: 10.1371/journal.pone.0216547.
[41] Cooksey-Stowers K, Schwartz MB, and Brownell KD. Food swamps predict
obesity rates better than food deserts in the United States. Int J Environ Res
Public Health. 2017; 14: 1366. doi:10.3390/ijerph14111366.
[42] NIMH. Obesity Hypoventilation Syndrome. Bethesda, MD: NIMH (2013).
90 The Internet of medical things

[43] CDC. Defining Adult Overweight and Obesity. Atlanta, GA: CDC
(2020).
[44] Petrilli CM, Jones S, Yang J, et al. Factors associated with hospitalization
and critical illness among 4,103 patients with COVID-19 disease in New
work City. medRxiv. 2020: 1–25. doi:10.1101/2020.04.08.20057794.
[45] Simonnet A, Chetboun M, Poissy J, et al. High prevalence of obesity in
severe acute respiratory Syndrome Coronavirus-2 (SARS-CoV-2) requiring
invasive mechanical ventilation. Obesity (Silver Spring). 2020; 28: 1195–9.
doi:10.1002/oby.22831.
[46] Firebaugh G and Acciai F. For blacks in America, the gap in neighbourhood
poverty has declined faster than segregation. Proc Natl Acad Sci USA. 2016;
113: 13372–7. doi:10.1073/pnas.1607220113.
[47] Nania R. Blacks, Hispanics Hit Harder by the Coronavirus, Early U.S. Data
Show. Washington, DC: AARP (2020).
[48] APM Research Lab. The Color of Coronavirus: COVID-19 Deaths by Race
and Ethnicity in the U.S. St. Paul, MN: APM Research Lab (2020).
[49] Kawachi I, Kennedy BP, Lochner K, and Prothrow-Stith D. Social capital,
income inequality, and mortality. Am J Public Health. 1997; 87: 1491–8.
doi:10.2105/AJPH.87.9.1491.
[50] Finch BK and Vega WA. Acculturation stress, social support, and self-rated
health among Latinos in California. J Immigr Health. 2003; 5: 109–17.
doi:10.1023/A:1023987717921.
[51] ODPHP. Access to foods that support healthy eating patterns. In: Healthy
People 2020. Rockville, MD: U.S. Department of Health and Human
Services (2020).
[52] Wu X, Rachek C, Sabath BM, Braun D, and Dominici F. Exposure to air
pollution and COVID-19 mortality in the United States: a nationwide cross-
sectional study. medRxiv. 2020: 1–36. doi:10.1101/2020.04.05.20054502.
[53] ODPHP. High school graduation. In: Healthy People 2020. Rockville, MD:
U.S. Department of Health and Human Services (2020).
[54] Shafa Z. New effort aims to provide Covid-19 resources to non-English
speakers in U.S. STAT News (2020).
[55] Lopez MH, Rainie L, and Budiman A. Financial and health impacts of
COVID-19 vary widely by race and ethnicity (2020, May 5).
[56] iSPARC. Coronavirus economic stimulus payments: who gets it, how, & impact
on other benefits. Psychiatry Inf Brief. 2020; 17: 1–4. doi: 10.7191/pib.1148.
[57] Abrams EJ, Myer L, Rosenfield A, and El-Sadr WM. Prevention of mother-
to-child transmission services as a gateway to family-based human immu-
nodeficiency virus care and treatment in resource-limited settings: rationale
and international experiences. Am J Obstet Gynecol. 2007; 197(3 Suppl):
S101–6. doi: 10.1016/j.ajog.2007.03.068.
[58] Kloppenburg K. Governor Little issues stay-at-home order for Idaho (2020,
March 25). Available online at: https://www.krtv.com/news/coronavirus/
stay-at-homeorder-issued-in-idaho.
 Impact of world pandemic “COVID-19” 91

[59] Orecchio-Egresitz RMISH. Georgia allowed some businesses to reopen today,

but many store and restaurant owners aren’t ready to take the risk (2020, April
24). Available online at: https://www.businessinsider.in/international/news/
georgia-allowed-some-businesses-to-reopen-today-but-many-store-andres-
taurant-owners-arent-ready-to-take-the-risk/articleshow/75365048.cms.
[60] Brown R and Scholl J. Idaho governor extends stay-home order through
April 30 because of coronavirus (2020, April 15). Available online at:
https://www.idahostatesman.com/news/coronavirus/article242012651.html.
[61] Chattu KV, Adisesh A, and Yaya S. Canada’s role in strengthening global
health security during the COVID-19 pandemic. Global Health Res Policy.
2020; 5: 16–18.
[62] Bhuyan HK, Kamila NK, and Pani SK. Individual privacy in data mining using
fuzzy optimization. Engineering Optimization, July, 2021 (Early Published).
[63] Bhuyan HK, Dash SK, Roy S. and Swain DK. Privacy preservation with penalty
in decentralized network using multiparty computation. International Journal of
Advancements in Computing Technology (IJACT). 2012; 4(1): 297–303.
[64] United Nations Development Program in China Assessment report on
impact of covid19 pandemic on Chinese enterprises. Available at: https://
www.cn.undp.org/content/china/zh/home/library/crisis_prevention_and_recov-
ery/assessment-report-on-impact-of-covid-19-pandemic-onchineseente.html
(accessed April 7, 2020).
[65] Wang X and Sun T. China’s engagement in global health governance: a
critical analysis of China’s assistance to the health sector of Africa. J Glob
Health. 2014; 4(1): 793–804.
[66] Wang Y, Hong A, Li X, and Gao J. Marketing innovations during a global crisis:
a study of China firms’ response to COVID-19. J Bus Res. 2020; 116: 214–20.
[67] Bhuyan HK, Raghu Kumar L, Reddy KR. Optimization model for sub-fea-
ture selection in data mining. 2nd International Conference on Smart
Systems and Inventive Technology (ICSSIT 2019), IEEE Explore, 2019.
[68] Bhuyan, HK and Reddy CV. Madhusudan. Sub-feature selection for novel
classification, International Conference on Inventive Communication and
Computational Technologies (ICICCT), IEEE Explore, 20–21 April, 2018.
doi: 10.1109/ICICCT.2018.8473206.
[69] World Health Organization. Coronavirus Disease (COVID-19) Update. World
Health Organization (2020). https://www.who.int/bangladesh/emergencies/
coronavirus-disease-(covid-19)-update (accessed November 7, 2020).
[70] BBS. Labour force survey Bangladesh 2016–17 (2018).
[71] Farhana KM and Mannan KA. Knowledge and perception towards Novel
Coronavirus (COVID 19) in Knowledge and perception towards Novel
Coronavirus (COVID 19) in Bangladesh. Int Res J Bus Soc Sci. 2020; 6: 76–9.
[72] Cousins S. Bangladesh’s COVID-19 testing criticised. Lancet (London,
England). 2020; 396: 591. https://doi.org/10.1016/S0140-6736(20)31819-5.
[73] World Health Organization. Increased Testing Capacity, Essential Step in
Fighting COVID-19. World Health Organization (2020). https://www.who.int/
92 The Internet of medical things

bangladesh/news/detail/08-10-2020-increased-testing-capacityessential-step-
in-fighting-covid-19 (accessed November 3, 2020).
[74] Bangladesh sees 100th death of doctors from Covid-19. Dhaka Tribune (2020).
https://www.dhakatribune.com/health/coronavirus/2020/10/15/bangladesh-sees-
100th-death-of-doctors-from-covid-19.
[75] Islam R. The impact of COVID-19 on employment in Bangladesh: Pathway
to an inclusive and sustainable recovery (2020).
[76] Sen B. Poverty in the time of Corona: Short-term effects of economic
slowdown and policy responses through social protection (2020).
[77] Sultan M, Hossain MS, Islam MS, Chowdhury K, Naim J, and Huq F.
COVID-19 impact on RMG sector and the financial stimulus package: Trade
union responses (2020).
[78] Ferguson NM, Laydon D, Nedjati-Gilani G, et al. Impact of non-
pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and
healthcare demand (2020). https://doi.org/10.25561/77482.
[79] World Population Prospects 2019, https://population.un.org/wpp/.
[80] Reuters. Bangladesh shuts down villages after tens of thousands attend cle-
ric’s funeral. New York Times (2020).
[81] Bradbury-Jones C and Isham L. The pandemic paradox: the consequences of
COVID-19 on domestic violence. J Clin Nurs. 2020; 19: 1–3. https://doi.org/
10.1111/jocn.15296.
[82] Islam A. COVID-19 Lockdown Increases Domestic Violence in Bangladesh.
DW (2020) https://www.dw.com/en/covid-19-lockdown-increases-domes-
tic-violence-inbangladesh/a-53411507.
 Chapter 5
Artificial intelligence in healthcare
Deepa Joshi1 and Anikait Sabharwal1

The ability of artificial intelligence (AI) to imitate human cognitive capabilities,

coupled with the ease of accessibility of medical data and the expeditious
advancement of analytical techniques, is bringing paramount difference to the
healthcare industry. Former research demonstrates the remarkable accuracy of AI
to aid physicians to settle on better clinical choices or supplant judgment made by
human beings, in particular, technical and practical areas of medical care. Statistics
indicate that the shortage of doctors in therapeutically under-resourced regions and
the lack of availability of skilled physicians in highly engaged clinical settings tend
to cause a rise in false detection rates. The excessive workload causes fatigue that
could lead to poor recovery of diseases. AI is ever-evolving as it is making
advancements at an exponential rate, especially in healthcare treatments, such as
monitoring treatments, improving the planning, and analyzing data to provide
better treatment plans. The procedure of data accumulation, data management,
clustering, and tagging invites numerous governance and regulatory challenges that
could take a long duration. The healthcare industry thrives in this unending battle as
the complexity and intricacy of data, and strict guidelines take a major toll. A
healthcare institute is subjected to ask for consent from an institutional review
board’s work to attenuate a portion of these concerns, and researchers and scientists
may measure, process, and anonymize DICOM information to strip away any
patient’s medical information. The AI-enabled medical care employed in the
institutes plays an imperative role as an informative assistant that provides aid to
doctors in acquiring an understanding of meaningful patterns from data collection.
This practice holds the potential to save a lot of time, cost, and effort while yielding
consistent, unbiased, and prime diagnosis or treatment. AI and deep learning (DL)
tools used in day-to-day medical decision-making have a grave impact on
improving the patient’s treatment and overall cost incurred due to their employ-
ment efficiency and accuracy. AI and machine learning (ML) can prove to be of
supreme importance and assistance in the early detection and thus the prevention of
a plethora of diseases by reading and analyzing the patient’s vitals. The presented
chapter is of 5-folds. First, the distinct data sources where the healthcare data is

1
School of Computer Science, University of Petroleum and Energy Studies (UPES), Dehradun, India
94 The Internet of medical things

gathered from are discussed. Second, we talk about the moral and lawful difficulties
of AI-driven medical care. Next, the structured and unstructured types of healthcare
data have been analyzed followed by the AI techniques applied to these types of data,
such as ML, natural language processing (NLP), and DL. We then analyze the crucial
disease areas such as cardiology, radiology, neurology, and cancer, where the health
issues can be alleviated by employing AI techniques. In conclusion, we further dis-
cuss the areas where AI-based techniques are applied in real life.

5.1 Introduction
The goal of AI is to teach, while also passing on human intelligence and instincts to
a computer. An amalgamation of mathematics, statistics, and science makes this
task attainable.
Famous AI strategies comprise ML techniques for structured data, for example,
neural network (NN), the traditional support vector machine (SVM), and advanced
DL, in addition to NLP that is utilized for the unstructured data. Cardiology, cancer,
and neurology are some of the principal disease areas that employ AI tools. Instead of
teaching, training and programming systems for certain tasks, ML emphasizes on
passing on the abilities concurrently. ML designs algorithms through which systems
can create and work on predictions for homogeneous situations and tasks. Under the
approach termed supervised learning, the machines are given data and are trained on
examples that have similar desired output so they can further work on unseen pro-
blems and data. Under unsupervised learning, just the input data is given to the
machine, and it is required to build an output without help and oversight [1]. From
the early days itself, ML algorithms were devised and were utilized to analyze
medical datasets. Today, ML is an instrumental base for providing key tools to astute
data analysis. Particularly over the most recent couple of years, the evolution of
technology gave moderately reasonable and accessible ways to gather and store data.
Current medical clinics are exceptionally provisioned with observing and other data
accumulation tools. Furthermore, the data is assembled and apportioned in huge data
frameworks. ML is right now at such an innovative stage that it is equipped to
analyze clinical data, and specifically there is a huge load of work done in clinical
analysis in compact dedicated diagnostic problems [2].
ML is being employed on varied kinds of diseases like diabetes and cancer,
falling under healthcare analysis. Cancer being one of the most fatal diseases has
distinct kinds that comprise breast cancer, stomach cancer, lung cancer, prostate
cancer, and the list goes on. Nearly 12% instances of cellular breakdown in the lungs
come every year, where 10% cases do not make it through. Likewise, for breast
cancer, 11% cases arise, of which 9% do not make it. For dealing with cancer in the
medical field analysis, it is crucial that accurate and prime quality data is produced.
In such an ambition-driven world, medical services should utilize the data in a par-
ticular way that consistently results in an ascent in the quality of medical care and a
significant decrease in the expense required for the therapy [3]. With time, especially
in recent years, there is an exponential rise in medical care research with ML. The
 Artificial intelligence in healthcare 95

difficulty human beings face in making decisions and inferring data is due to the
assortment of medical data that involves omics data, clinical data, or even EHR data.
For that very reason, ML has been effectively presented and advanced in medical
care for finer comprehension of data and superior dynamic cycle [4].
The chapter has been presented in 5-folds. Initially, the distinct data sources
from where the medical care data can be accumulated are examined. Second, we
discuss the moral and legal hurdles of AI-driven healthcare. Furthermore, the
structured and unstructured kinds of medical services data have been investigated,
which is trailed by the AI strategies applied to these kinds of data, for instance, ML,
DL, and NLP. We at that point dissect the absolutely critical disease areas like
cardiology, radiology, nervous system science, and cancer, where the medical
problems can be lightened by utilizing and efficaciously employing AI methods. In
conclusion, we examine the fields where AI-based methods are applied in reality.

5.2 Healthcare data sources

If not handled with utter perseverance and seriousness, data accumulation and
hence management in healthcare can be particularly perplexing, laborious, and
unsafe. In the light of staying aware of this consistently converting data landscape,
the associations need a data accumulation solution that is prepared to deal with
medical care, crystal clear use cases and with safety and security efforts, and
compliance that consider the novel requirements of the industry.
The healthcare industry has a plethora of sources for big data that include the
records of hospitals, medical records of patients, medical examination results, and
IoT (Internet of Things) devices as illustrated in Figure 5.1. A notable segment of

Examination

Inpatient health Patient-generated

monitoring health data

Medical Laboratory
imaging results

Figure 5.1 Various sources of healthcare data [37]

96 The Internet of medical things

big data that is instrumental in public healthcare is generated by biomedical

research. Thus, this portion requires careful management and analysis to derive
meaningful information. Over the years, the analysis of healthcare data is coming
up to become quite possibly the majority of favorable research sectors. It comprises
information for varied types, like sensor data, clinical data, and omics data. The
patient records that are stored in electronic health records during the ongoing
treatment are termed clinical data. Omics data is amongst the high magnitude data
that comprises transcriptome, proteome, and genome data types. Sensor data is the
one that is gathered from numerous wearable and wireless sensor devices [4].
Each industry faces challenges regarding regulations and governance in the
collection, storage, and management of data, although when it comes to the
healthcare industry, the battles fought are one of a kind. Frequently, the information
that is gathered in medical care is intricate and is administered by unnecessarily
rigid guidelines [5]. The majority of healthcare providers collect data in innumer-
able forms, like forms for patient intake, treatment evaluation, health assessment
forms, and consent forms amongst many. To make the information attainable in the
customer relationship management systems at the backend, electronic well-being
record frameworks, and divided between suppliers, the data post collection is made
to go through manual data entry. Statistics show that this manual entry process is
prone to errors like incorrect entries, typing errors, or even wrong form entries
being made in a patient’s records. Usually, the forms on paper render patient
information unavailable to all custodians inside 24 h, particularly the data gathered
from outside sources. This may bring the patient a disconnected care environment.
The most genuine kind of digital clinical information is acquired right at the mark
of care at a clinical office, emergency clinic, hospital, or practice. Regularly
referred to as the electronic clinical record (EMR), the EMR is typically unavail-
able to people beyond the research spectrum. The data accumulated involves reg-
ulatory and demographic data, conclusion, treatment, physician endorsed drugs,
research facility tests, physiologic observing data, hospitalization, patient insur-
ance, and so on. Individual associations, for instance, medical clinics or health
frameworks offer access to the internal staff, although bigger coordinated efforts,
for instance, the NIH Collaboratory Distributed Research Network, give multi-
disciplinary access to clinical information vaults by qualified researchers.

5.3 Legal and ethical obstacles of artificial intelligence-

driven healthcare
In India, the data transfer for processing data is not covered by any definite law.
Gigantic measure of information was handled by a third party by the ones providing
service, hence recording the data as indicated by US laws [6]. A debate struck in
2017 because of a comparable arrangement among Google DeepMind and the
Royal Free London NHS Foundation Trust [7]. That understanding was repri-
manded for not adhering to the Caldicott Principles by moving information that is
more than what is needed and obscuring the complete line that is between the data
 Artificial intelligence in healthcare 97

regulators and the ones who process data. This led to legitimate commitments and
liabilities. The United Kingdom is known to have instrumental mechanisms like the
Information Commissioner’s Office (answerable for authorizing as well as putting
Data Protection Act into motion) and Health Research Authority (liable for a
government foundation diving into health research) and Confidentiality Advisory
Group (a strategy for classified health data in if explicit assent is unavailable),
which were not consulted prior to the rise of data transfer, subsequently done to
employ “self-assessment information governance toolkit” utilized to support the
security of particularly specialized technical framework to deal with NHS data [7]. In
such cases, it is crucial that the care-providers are cautious when they share the data
with an outsider which is in an implicit relationship with the separate patient(s),
respectively. The care that is provided straight to the concerned authority or person is
regarded as “action concerned about the protection, examination and treatment of
disease and the mitigation of enduring of a recognized individual.” This requires a
notice to be provided to the patient or subject of concern. In case the explicit assent
and notice are not granted, then all the data, irrespective of being tagged or untagged,
falls into the public domain and is required to be produced by a statutory body.
Through this, a tab can be kept regarding the unlawful proprietary manipulation of the
information and accordingly the data processor can only exchange the data up to a
certain limit. The freely accessible public or peer-examined data sets, for example,
Messidor, will help unconstrained development of processes and algorithms, since
without assent such de-identified data sets ought to be viewed as a community
resource. The idea of putting them in the hands of the community is to employ and
enforce policing of the data exchanges at a level that cannot be attained by the gov-
ernment. In India, we need to have legal regulations, for example, segment 251 in the
United Kingdom, which calls the government and legal control for such exchanges
[7]. As a matter of fact, the discussion of ownership and custodial responsibility of
such information in our country seldom takes place [8]. There is no particular brief
regarding what way the information will be handled by the machine in the event that
we utilize an AI device on such sets. The effect of transfer learning and AGI is such
that humans often are boggled as to how the data is handled by the machine. People
apparently found certain lines of play in AlphaZero “completely alien” but highly
efficacious. Notable progress has been seen in Justice BN Srikrishna Committee, as
they strategically empowered their patients [8]. In any case, it is reckless to anticipate
that general protection should address immensely tangled bioethical worries in the
expansion and growth of clinical AI. The healthcare profession is required to demand
explicit customs and “Dos and Don’ts,” which whenever stuck to will guard it against
prosecution and litigation in the event of data break since India is not just the colossal
producer and the least expensive source of such data (as per confirmation made by the
authors in the Google Diabetic Retinopathy Project) [9], it will pose to be a massive
market specifically for the algorithms that are derived from it in future. The machines
require rules and regulations for effective recursive self-improvement, which calls for
a dire need of strengthened bioethical and computational ethics framework to guar-
antee optimum and safe functioning of the same. The phrase, “First, do no harm,”
should be strongly conveyed from clinical morals into the discipline of computational
98 The Internet of medical things

bioethics. We possess the characteristics of a typical parent in this case; past a phase
of development, we may not be in charge of these algorithms any further or may not
comprehend them by any stretch of our imagination.

5.4 Types of healthcare data

Clinical data is the key resource when it comes to carrying out research in medical
and health research work. Health information can be located in numerous struc-
tures, as vital signs, lab results, patient’s way of life information, notes by the
doctor, and an assortment of kinds of imagery (ultrasonography, ultrasonography,
and magnetic resonance imaging, pathology slides). Despite there being no prin-
ciples or classifications that include a wide range of medical information, it is
extremely useful to think about this significant data as organized and unstructured
information. Getting a handle on and accommodating the two prime kinds of data,
organized, structured, and unstructured data, is among indispensable difficulties for
medical care suppliers. The clinical data, which is rising dramatically, incorporates
an incredible volume of unstructured and structured information as different areas
[10]. The grave issue lying in medical care fields is that about 80% of information
for electronic medical records or a majority of clinic data frameworks are in general
disregarded, unsaved, or even deserted in a lot of clinical places for quite a while
[4]. However, the entire data is as yet made in numerous hospitals quite difficult to
be associated with clinical research of big data and the AI industry in medical
services. Subsequently, there is a dire need to deal with those unmanaged data.

5.4.1 Structured data

Structured data is a type that is kept in store in an utterly organized way, like
documents. It is simpler to analyze and hence store since it has direct limits and is
produced and put away in a standard design and format. Patient segment data,
analysis and interaction codes, medicine codes, and other explicit information from
the electronic medical record are ordinarily created and delivered in a standard,
structured organized way. Traditional data warehouses can be considered the ones
that are quite frequently structured data.
Unstructured data is rather a bit laborious. It comes in a plethora of forms that
include emails, audio and video files, text and genome files, as well as social media
posts. Structured data is certainly not an analogous category—if the data is organized
and structured, it is not necessary that it is organized in a manner that makes com-
plete sense or has ease of interpretation. Contrarily, if the data lacks formal structure,
it is not necessary that it will not be interpreted with absolute ease, or that it can only
be analyzed by choosing a resource-intensive manner. On the face of it, unstructured
data addresses a more prominent hurdle to dissect and decipher than structured data.

5.4.2 Unstructured data

Unlike structured data, unstructured data is not entirely defined and thus cannot be
inspected the way structured data is. Therefore, it is rather difficult for healthcare
 Artificial intelligence in healthcare 99

organizations to make unstructured data operable. Enormous amounts of unstruc-

tured healthcare data are collected by healthcare organizations, and quite a few
difficulties to discover the instruments that permit that data to be utilized.
Unstructured data can be made pragmatic with AI and ML, hence enabling them to
help doctors gather instrumental insight into the patients’ health [10]. Unstructured
big data in medical systems prior to proposing the development of healthcare AI, at
the moment, has its base on ML. To enhance the utilization and application of
unstructured healthcare big data, we need to establish the data accumulation,
quality assurance, and anonymization processes. It is essential that metadata for
every type of unstructured healthcare data is extracted, defined, visualized, and
standardized automatically. Further, the open stage for incorporation and the
application of the unstructured medical data should be created while mirroring
these ideas. A cautious glance at this dichotomy, particularly inside the setting of
evolving innovation, uncovers nuanced contrast.
Images and free text are not that easy to categorize in a way that numerical,
structured data points can be categorized. For example, diagnosing the blood
pressure reading as elevated, normal, or hypertensive can be achieved merely by
some lines of code. A doctor’s note that indicates “gen fatigue, trouble breath, and
chest pain” indicates hypertension, although abbreviation and spelling mistakes
made in text would need a human being to decode and further understand it (mainly
when the text is written by hand or is manually scanned in the EMR through fax).
Imagery poses difficulties in the same line—X-rays and pathology slides for the
most part are incomprehensible to each aside from exceptionally skilled experts
and even doctors who have years of experience often further require second judg-
ment in order to confirm an inference or diagnosis. In contrast, the time when
clinical imaging is progressively depending upon digital imagery, the unstructured
data is thus predominantly examined and analyzed by hand. However, persistent
progresses in ML and AI can possibly change the manner in which doctors and
suppliers utilize unstructured data. As per the example provided, a tool of NLP may
have the ability to understand a note made by the doctor and understand the same as
“general fatigue, trouble breathing, and chest pain” whereas a tool for ML decision
support may have the ability to recommend that these are manifestations identified
with hypertension (this analysis likewise benefits from structured context-oriented
information like the patient’s stature, pulse, and weight). In the same manner, using
huge repositories of medical imagery, computer science researchers are con-
scientiously working with doctors in order to train and build ML models to perceive
patterns in medical imagery to give a computerized and automated “subsequent
assessment” affirming (or proposing ambiguity) a manually created inference or
finding [11].
AlphaZero is known to have become proficient in chess, Shogi and Go unac-
companied with human assistance, apart from the rules of the game. In a period of
24 h, AlphaZero triumphed over all the exceptional AI programs like AlphaGo,
Elmo, and Stockfish (3 days) [9]. Improvement in AI is presently moving its
attention from “supervised” learning (which needed numerous labeled instances to
prepare the instrument to perceive comparable patterns) to “unsupervised learning”
100 The Internet of medical things

(a type of learning in which the machine trains without the help of any labeled
data). With time, progress, having such an eminent computational force as its
backbone, AI is becoming supreme. Although considering this, it raises a major
concern about a situation with this force discovering its way into unacceptable and
dangerous hands, be it artificial or human. The earlier one evolves at the pace of
human advancement, giving an open door to evolve and reclaim our lives, however,
not so for the artificial elements, as we can see with the AlphaGo experience. This
calls for deeper caution in the fields of medicine and medical research since the
individual influenced by every choice is a conscious human being. AI cannot be
considered one single technology, but rather an assembly of them, and many of
them have a direct, yet distinct, relationship with the field of healthcare.

5.5 AI techniques developed for structured and

unstructured data
AI, which comprises robotics, NLP, and ML may be employed in nearly all kinds
of fields in healthcare, and the possible commitments to clinical education, bio-
medical research, and the applications of medical services appear to be boundless.
With its thriving capacity to coordinate, amalgamate, and grasp from massive sets
of medical information, AI serves parts in clinical decision-making, diagnosis, and
customized medication. Symptomatic and diagnostic calculations based on AI
integrated with mammograms are aiding the discovery and diagnosis of breast
cancer, filling in as second judgment for radiologists. Furthermore, progressed
virtual symbols are equipped for participating in significant discussions, which has
effective inferences for the determination and treatment of disease related to psy-
chiatry. Applications in AI additionally reach out to the actual physical domain
with physical task support networks and systems, robotic prostheses, and
portable controllers aiding the delivery of telemedicine.
Devices in AI can be essentially categorized into two distinct parts. The initial
classification incorporates ML methods that investigate structured data like EP,
hereditary, and imaging information. In clinical implementation, ML systems
cluster patient’s attributes and traits or construe the likelihood of the outcomes of
the diseases. The subsequent classification incorporates NLP techniques that gather
from unstructured data’s information like medical journals or notes to enhance and
improve medical structured data. The NLP systems aim at transforming texts to
machine-decipherable structured data, which is then analyzed by ML techniques.

5.5.1 Machine learning

With the aim of extracting features from data, ML builds various data analytical
algorithms. Usually patient “traits” and medical outcomes of interests are taken as
inputs for the ML algorithms. These traits often comprise the standard data such as
gender, age, and history of diseases, along with data that is specific to diseases, that
include EP test, gene expressions, clinical symptoms, medication, diagnostic ima-
ging, physical examination results, and the list goes on. Not just the characteristics,
 Artificial intelligence in healthcare 101

even the patient’s medical outcomes that comprise quantitative disease levels,
patient’s survival times, and disease indicators, like, size of tumor, are gathered for
the purpose of clinical research [12]. The subsequent features are preferred to
accelerate ML system’s effectiveness in resolving diagnostic tasks in healthcare:
great execution, the capacity to suitably manage noisy data (errors) and with
missing data, the clarity of knowledge of diagnosis, the capacity to elaborate con-
clusions, and the capacity of the algorithm to reduce the quantity of examinations
that are important to acquire reliable diagnosis [2]. ML algorithms may be sepa-
rated in three significant classes: supervised learning, semi-supervised learning,
and unsupervised learning. The one that is immensely notable for feature extraction
is unsupervised learning, while supervised learning is reasonable for prescient
modeling through progressively developing a few connections between the
patient’s characteristics (input) in addition to the interest’s outcome (output). The
crossbreed between supervised learning and unsupervised learning has been
effectively put forward as semi-supervised learning that is appropriate for situations
where the outcome is missing for particular subjects [12]. Figure 5.2 shows the
ubiquity of distinct supervised learning strategies in medical implementations.
AI applications in healthcare typically employ supervised learning for a ple-
thora of reasons and purposes. Unsupervised learning may be utilized as a feature
of the step of preprocessing to lessen dimensionality or recognize subgroups, which
therefore makes the subsequent supervised learning’s step way more effective.
Important strategies incorporate decision tree, random forest, naı̈ve Bayes, dis-
criminant analysis, NN, logistic regression, linear regression, nearest neighbor, and
SVM [13].

5.5.2 Neural networks

NNs, when regarded with logistic regression, may be considered an add-on to
encapsulate composite nonlinear relationships between an outcome and the input

Support vector machine

Neural network
Logistic regression
Discriminant analysis
Random forest
Linera regression
Naïve bayes
Nearest neighbor
Decision tree
Hidden markov
Others

Figure 5.2 The ML algorithms employed in the healthcare literature. Data is

created by looking through the ML algorithms inside medical services
on PubMed [12]
102 The Internet of medical things

variables. The methodology of NNs in tackling such issues is not quite the same as
typical approaches. It takes the countless contributions of many such manually
written digits named as training samples and builds a framework to naturally derive
rules for distinguishing the digits by persistently learning patterns from those
training samples. Moreover, the precision of such models can be further refined by
appending additional training samples [13]. “Perceptron” is the prime belief that
drives the building of an NN. The concept of a perceptron was encouraged by the
human nervous system, where each network consists of distinct interconnected
neurons that effectively communicate with one another.
Over the period of time, healthcare has become an instrumental field where AI
techniques, including DL, are extensively employed and applied. DL has the ability
to inspect more mind-boggling nonlinear patterns in the data. With the evolving
technology, there is a grave rise in the amount and complexity of data, which calls
for the new fame of DL. An algorithm that has its base on DL functions in the
subsequent manner: input layer is utilized as a gateway for the input data to enter
the deep NN. Figure 5.3 mentions the sources of data for DL.
The design of the input data gets changed so the deep NN realizes the way to
handle and process the same, and potentially certain extra calculations are then
performed. At that point, the outcomes are provided to the initial hidden layer. A
network can contain a variety of secret layers, implying that this footstep carrying
out a calculation and providing the result to the following hidden layer is further
carried out on various occasions, until and unless the yield layer has been reached.
The final layer’s role is essential to offer a solution to the question that DL is
expected to answer.
For instance, in Figure 5.4, a prostate magnetic resonance (MR) picture is the
input that goes via the network with every one of its layers, and thus the output
layer conveys the response to the doubt “Is there a lesion in the prostate MR
image?”

5.5.3 Natural language processing

The EP, picture, and hereditary data can be deciphered by the machine with the
goal that the ML algorithms can be instantly performed after legitimate pre-
processing or quality maintenance measures. In any case, an immense extent of

Diagnostic imaging
Electrodiagnosis
Genetic diagnosis
Clinical laboratory
Mass screening
Others

Figure 5.3 The sources of data for deep learning. The data is created by looking
through deep learning in an amalgamation with the diagnosis methods
on PubMed [38]
 Artificial intelligence in healthcare 103

This MRI prostate

image does
contain a lesion

This MRI prostate

Does this image does not
prostate MR contain a lesion
image contain Etc.
a lesion? Input layer Hidden Output

Figure 5.4 A deep neural network utilizes various layers to protect and handle
input data (a prostate MR picture, in this case). Once the data goes via
the network, a solution is given by the algorithm to a particular
question, like, “Is there a lesion in the prostate MR image?” [38]

clinical data is in the structure of narrative text, like physical inspection and
assessment, clinical research center and lab reports, and employable notes that are
unstructured while possessing no limits for the PC program. With regard to this
unique circumstance, NLP focuses on separating and hence gathering helpful data
through narrative text in order to help medical decision-making. Two essential
components can be found in an NLP pipeline, namely (1) processing of text and (2)
subsequent classification. It recognizes a progression of catchphrases and keywords
related to diseases in the clinical notes dependent on the verifiable historical
databases through text processing [14]. Following that, a proportion of the key-
words are picked by analyzing their impacts on the categorization of the abnormal
and normal cases. At that point, the approved keywords advance the structured data
to help clinical dynamics of making decisions. The development of the NLP
pipelines is with a goal of aiding clinical making of decisions to notify arrange-
ments related to treatment, overseeing unfavorable effects, and much more. For
instance, Fiszman et al. presented that bringing NLP into the picture for perusing
the chest X-beam reports provides aid to antibiotic associate frameworks to caution
doctors for the conceivable requirement against infective therapy [15]. On a similar
note, NLP was employed by Miller et al. to oversee the lab-based antagonistic
effects on its own [16]. NLP pipelines are also instrumental in assisting diagnosis of
diseases. For example, Castro et al. recognized 14 disease factors related to cere-
bral aneurysms by carrying out NLP on the medical and clinical notes [17].

5.6 Major disease areas

AI, which incorporates the domains of NLP, robotics, and ML, may be imple-
mented in practically any domain in healthcare, and its possible commitments to
medical education, clinical research, and applications of medical services appear to
be boundless. Flaunting its hearty capacity to coordinate, learn, and integrate from
104 The Internet of medical things

huge datasets of the medical industry, AI has the potential to serve parts in analysis,
diagnosis [18] customized medication [19], and decisions made in the healthcare
industry [20]. For instance, diagnostic calculations and applications based on AI
applied to mammograms are aiding the identification of breast cancer, filling in the
second judgment cum opinion for radiologists [21]. Likewise, progressed virtual
symbols are equipped to take part in significant conversations, having inferences
for the diagnosis and thus, treatment of diseases in psychiatry [22]. AI imple-
mentations additionally reach out into the physical sphere with mobile manip-
ulators, physical task support systems, and robotic automated prostheses, and
aiding the conveyance of telemedicine.
Cancer, neurology, and cardiology are few of the key disease areas that employ
AI tools [12]. Somashekhar et al. proposed that the IBM Watson for oncology must
be a top-notch AI system for aiding the diagnosis and analysis of cancer via a two-
staged approval study [23]. Esteva et al. carefully performed an analysis on clinical
pictures to recognize subtypes of skin malignant growth [24]. Bouton et al.
developed an AI structure to restore and subsequently reenforce the control of
development in patients with quadriplegia [25]. Farina et al. effectively tried the
power of an offline machine or man-made interface that uses the issued timings of
spinal motor neurons to control the prostheses of the upper limb [26]. Dilsizian and
Siegel inspected the normal utilization of the AI system to analyze and examine the
anticipated utilization of the AI system heart disease via a careful look at the car-
diovascular pictures [19].

5.7 Applications of AI in healthcare system

Application of AI knows no bounds, as it proves to offer immense advancements in
each of the domains of medicine, ranging from the diagnosis to the entire treatment.
The credit to on-demand healthcare services goes entirely to the advances made in
smartphones and wireless technology that have been made using search engines or
platforms and apps that track health possible, while also enabling a new form of
healthcare delivery accessible whenever and wherever required, via remote inter-
actions. These assistance services are beneficial for underprivileged places and
areas that lack specialists by helping bring the cost down and preventing pointless
exposure to infectious diseases at the center. An absolute blend of bigger data
accumulation data libraries, a large AI pool, and accelerated computer processing
speed has featured quick building of AI technology and tools healthcare [27].
Diving into specifics, the entire advancement of DL has made a major impression
on the lines we perceive AI tools these days and is practically the cause for the
majority of recent exhilaration and anticipation surrounding AI applications
(Figure 5.5). A plethora of favorable implementations for medical purposes have
varied opinions. Forbes issued a statement in 2018 that the instrumental domains
would be image analysis, clinical decision support, virtual assistants, robotic sur-
gery, and administrative workflows [28]. A 2018 written report by Accenture
referred to the particular areas while also including dosage error reduction,
 Artificial intelligence in healthcare 105

Application Potential annual value by 2026 Key drivers for adoption

Technological advances in robotic
Robot-assisted surgery $40B
solutions for more types of surgery
Increasing pressure caused
Virtual nursing assistants 20
by medical labor shortage
Easier integration with
Administrative workflow 18
existing technology infrastructure
Need to address incresingly complex
Fraud detection 17
service and payment fraud attempts
Prevalence of medical error,
Dosage error reduction 16
which leads to tangible penalties
Proliferation of connected
Connected machines 14
machines/devices
Patent cliff; plethora of data;
Clinical trial participation 13
outcomes-driven approach
Interoperability/data architecture
Preliminary diagnosis 5
to enhance accuracy
Storage capacity; greater
Automated image diagnosis 3
trust in AI technology
Increase in breaches;
Cybersecurity 2 pressure to protect health data

Figure 5.5 Applications of AI in the healthcare system [39]

connected machines, and cybersecurity [29]. A 2019 report written by McKinsey

mentions key areas being robotics-assisted surgery, electroceuticals, targeted and
personalized medicine, and connected and cognitive devices [30].

5.7.1 Solutions based on genetics

It is being counted on that within the coming decade a major portion of the
worldwide populace will be provided with a full genome sequencing either upon
the time of birth or in grown-up life. Such genome sequencing takes up approxi-
mately 100–150 GB of data and allows a prominent tool for accurate medicine. The
process of uniting the interface of the genomic and phenotypic information is
currently in progress. The present-day clinical system requires refinement and
redesign in order to utilize certain genomics data and its advantages [31].

5.7.2 Development and discovery of drug

Drug development and development is one such domain that is extensively long,
costly, and complex and holds the probability of taking greater than 10 years, right
from the recognition of molecular targets to the point unless a drug item is affirmed
and promoted in the market. Disappointment at this point in time poses an astounding
financial impact, and quite a few drug candidates fail amidst development and are not
able to make it to the market [32]. There are numerous methods that have been
utilized to develop models to aid drug discovery since the 1990s such as NNs, ran-
dom forest, and SVMs. Not long ago, DL started to be carried out because of the
expanded measure of information and the consistent enhancements in computing
power. ML can be employed in the drug discovery process for streamlining various
106 The Internet of medical things

tasks. They comprise de novo design of drug–receptor interactions, drug compound

property, drug reaction prediction, and activity prediction [33].

5.7.3 Support in clinical decisions

The help of NLP makes it more advantageous for doctors to narrow down the
pertinent data from patient reports. AI holds the capacity to store and work on huge
arrangements of information, which can give knowledge databases and encourage
assessment and recommendation independently for every patient, consequently
assisting clinical decision support [34].

5.7.4 Robotics and artificial intelligence-powered devices

There are various fields in medical care where one can see robots being utilized to
supplant the human workforce, increase human capacities, and help human
healthcare experts. These also comprise robots employed for surgeries, for exam-
ple, assistants for rehabilitation, patient assistance, and laparoscopic operations,
robots that are coordinated into prosthetic and implants, and robots employed to
help doctors and other medical care staff working on their routine errands. A por-
tion of these gadgets are being created by organizations particularly for interaction
with patients and, hence, advancing the association among people and machines
from a care perspective. Many of the robots right now being worked on have a fair
degree of AI innovation fused for better execution with regard to classifications,
language recognition, and image processing [34].

5.8 Examples of AI used in healthcare

Large health companies are now often merging, which allows for pronounced
health data accessibility [35]. Greater health data helps in laying the fundamental
groundwork for the implementation of AI algorithms. A huge portion of indus-
tries in the healthcare sector focus on the implementation of AI in the support
systems of clinical decision-making. ML algorithms adjust and take into con-
sideration more powerful feedback followed by the solutions as the collection of
data occurs [36]. Big data is becoming the talk of the town amongst a number of
companies to explore the possibilities of its incorporation in the healthcare
industry. The organizations look into the market opening via the domains of “data
storage, analysis, assessment, and management technologies,” the most important
aspects of the medical industry. Following are listed certain key examples of big
shot companies that have been instrumental in contributing to AI algorithms for
use in healthcare.
IBM’s Watson Oncology is building at the Memorial Sloan Kettering Cancer
Center and Cleveland Clinic. At the same time, IBM is actively functioning with
CVS Health on applications in AI in the treatment of chronic diseases, along with
Johnson on the analysis of scientific papers in order to locate associations for
advancement in development of drugs. In May 2017, Rensselaer Polytechnic
Institute and IBM started a project that they were jointly working on, granted
 Artificial intelligence in healthcare 107

Health Empowerment by Analytics, Learning and Semantics (HEALS), to prospect

employing technologies of AI to augment the medicine industry [23].
Microsoft’s Hanover project, partnered with Oregon Health Science
University’s Knight Cancer Institute, plays a paramount role in analyzing medical
research with the agenda of predicting the most efficacious alternatives for cancer
drug therapy in patients. There are various project initiatives that comprise devel-
opment of programmable cells along with medical image analysis of tumor
progression.
Google’s DeepMind platform is currently extensively employed by the UK
National Health Service on data accumulated through applications on mobile
phones to detect certain health risks. Another project under the NHS includes an
examination of clinical pictures gathered from NHS patients with an aim to develop
computer vision algorithms for spotting cancerous tissues.
Tencent at this point is actively operating on various clinical frameworks and
deliverables. The aforesaid compass AI Medical Innovation System, a diagnostic
medical imaging deliverable that is powered by AI; WeChat Intelligent Healthcare;
and the Tencent Doctorwork.
Intel’s venture capital arm, Intel Capital, a while back funded a startup,
Lumiata, that employs AI to develop care options and identify patients at risk.
Kheiron Medical built a DL software for analyzing mammograms and
detecting breast cancers in them.
Fractal analytics has incubated Qure.ai that pays extensive attention to
employing AI and DL to refine radiology and accelerate the diagnostic X-rays’
analysis.
Elon Musk launched the surgical robot that inserts the Neuralink
brainchip. Neuralink has emerged with trailblazing neuro-prosthetics that networks
with a huge number of neural pathways in the mind [35]. According to the procedure,
a chip, approximately the same size of a quarter, may be embedded instead as a piece
of skull precisely by a surgical robot that avoids accidental injury [27].
The Indian startup, Haptik, lately built a Chabot functioning in WhatsApp
that answers questions corresponding to the deadly coronavirus in India.
Digital consulting applications, including AliHealth Doctor You, Babylon
Health’s GP at Hand, Your.MD, Ada Health, and KareXpert, employ AI for clinical
consultancy dependent on the patient’s clinical history and basic clinical proficiency.
The consumers are required to fill in their symptoms into the application that then
uses speech recognition to collate against a disease database. Babylon further takes
the consumer’s medical history into account and presents a recommended action.
Entrepreneurs in the medical industry successfully utilize seven business model
archetypes in order to bring AI solutions to the market. These archetypes actively
hang on to the value that is produced for the target consumer (e.g., a patient’s focal
point versus medical care supplier’s focus) and take into careful consideration cap-
turing components (for instance, providing data or associating partners). iFlytek set in
motion a service robot “Xiao Man” that fused AI technology for identifying the
consumers registered and providing customized suggestions in medical areas. It also
functions in the domain of medical imaging. Companies like UBTECH (Cruzr) and
108 The Internet of medical things

SoftBank Robotics (Pepper) are also actively working toward building similar robots.
Far reaching tech companies such as Google, Amazon, Baidu, and Apple, all have
dedicated research divisions in AI along with having smaller AI-based companies
acquired worth millions of dollars since the field of AI is constantly evolving.
Numerous automobile manufacturers are thinking in the same line and are beginning
to employ ML medical care in their cars.

References
[1] D. Joshi, S. Anwarul and V. Mishra. Deep learning using Keras. In Machine
Learning and Deep Learning in Real-Time Applications; 2020 (pp. 33–60).
IGI Global. doi:10.4018/978-1-7998-3095-5.ch002.
[2] I. Kononenko. Machine learning for medical diagnosis: history, state of
the art and perspective. Artificial Intelligence in Medicine. 2001; 23(1):
89–109.
[3] K. Raheja, A. Dubey and R. Chawda. Data analysis and its importance in
health care. International Journal of Computer Trends and Technology.
2017; 48(4): 176–180.
[4] A. Dhillon and A. Singh. Machine learning in healthcare data analysis: a
survey. Journal of Biology and Today’s World. 2019; 8(6): 1–10.
[5] D. Joshi and T.P. Singh. A survey of fracture detection techniques in bone
x-ray images. Artificial Intelligence Review. 2020; 53(6): 4475–4517.
[6] J. Bali, R. Garg and R.T. Bali. Artificial intelligence (AI) in healthcare and
biomedical research: why a strong computational/AI bioethics framework is
required? Indian Journal of Ophthalmology. 2019; 67(1): 3.
[7] J. Powles and H. Hodson. Google DeepMind and healthcare in an age of
algorithms. Health and Technology. 2017; 7(4): 351–367.
[8] V. Khullar. White Paper on Data Protection Framework for India [Internet].
Prsindia.org; 2018. [cited 2018 August 25]. Available from: http://www.
prsindia.org/uploads/media/Report.
[9] J. Krause, V. Gulshan, E. Rahimy, et al. Grader variability and the impor-
tance of reference standards for evaluating machine learning models for
diabetic retinopathy. Ophthalmology. 2018; 125(8): 1264–1272.
[10] HIT Infrastructure. Unstructured Healthcare Data Needs Advanced Machine
Learning Tools; 2018. [online] Available at: https://hitinfrastructure.com/
news/unstructured-healthcare-data-needs-advanced-machine-learning-tools
[Accessed 23 March 2021].
[11] HealthSnap, Inc. Challenges and Opportunities of Structured and Unstructured
Health Data | HealthSnap INC; 2018. [online] Available at: https://healthsnap.
io/structured-unstructured-health-data/ [Accessed 23 March 2021].
[12] F. Jiang, Y. Jiang, H. Zhi, et al. Artificial intelligence in healthcare: past,
present and future. Stroke and Vascular Neurology. 2017; 2(4).
[13] I. Goodfellow, Y. Bengio and A. Courville. Deep learning. First Edition.
Cambridge, MA: The MIT Press; 2016.
 Artificial intelligence in healthcare 109

[14] C.D. Manning and H. Schütze. Foundations of Statistical Natural Language

Processing. Cambridge, MA: MIT Press, 1999.
[15] M. Fiszman, W.W. Chapman, D. Aronsky, et al. Automatic detection of
acute bacterial pneumonia from chest X-ray reports. Journal of the American
Medical Informatics Association. 2000; 7: 593–604.
[16] T.P. Miller, Y. Li, K.D. Getz, et al. Using electronic medical record data to
report laboratory adverse events. British Journal of Haematology. 2017;
177: 283–286.
[17] V.M. Castro, D. Dligach, S. Finan, et al. Large-scale identification of
patients with cerebral aneurysms using natural language processing.
Neurology. 2017; 88: 164–168.
[18] F. Amato, A. López, E.M. Peña-Méndez, P. Vaňhara, A. Hampl and
J. Havel. Artificial neural networks in medical diagnosis. Journal of Applied
Biomedicine. 2013; 11(2): 47–58.
[19] S.E. Dilsizian and E.L. Siegel. Artificial intelligence in medicine and cardiac
imaging: harnessing big data and advanced computing to provide persona-
lized medical diagnosis and treatment. Current Cardiology Reports. 2014;
16(1): 441.
[20] C.C. Bennett and K. Hauser. Artificial intelligence framework for simulating
clinical decision-making: a Markov decision process approach. Artificial
Intelligence in Medicine. 2013; 57(1): 9–19.
[21] J. Shiraishi, Q. Li, D. Appelbaum and K. Doi. Computer-aided diagnosis and
artificial intelligence in clinical imaging. Seminars in Nuclear Medicine.
2011; 41(6): 449–462.
[22] D.D. Luxton. Artificial intelligence in psychological practice: current and
future applications and implications. Professional Psychology: Research and
Practice. 2014; 45(5): 332–339.
[23] S.P. Somashekhar, R. Kumarc, A. Rauthan, et al. Abstract S6-07: double
blinded validation study to assess performance of IBM artificial intelligence
platform, Watson for oncology in comparison with Manipal multi-
disciplinary tumour board? First study of 638 breast Cancer cases. Cancer
Research. 2017; 77(4 Suppl): S6-07.
[24] A. Esteva, B. Kuprel, R.A. Novoa, et al. Dermatologist-level classification
of skin Cancer with deep neural networks. Nature. 2017; 542: 115–118. doi:
10.1038/nature21056.
[25] C.E. Bouton, A. Shaikhouni, N.V. Annetta, et al. Restoring cortical control
of functional movement in a human with quadriplegia. Nature. 2016; 533:
247–250.
[26] D. Farina, I. Vujaklija, M. Sartori, et al. Man/machine interface based on the
discharge timings of spinal motor neurons after targeted muscle reinnervation.
Nature Biomedical Engineering. 2017; 1: 0025. doi: 10.1038/s41551-016-0025.
[27] K.-F. Lee AI Superpowers: China, Silicon Valley, and the New World Order.
Boston, MA and New York: Houghton Mifflin; 2018.
[28] B. Marr . “How Is AI Used in Healthcare: 5 Powerful Real-World Examples
That Show the Latest Advances”. Forbes, July 27, 2018. Available at: https://
110 The Internet of medical things

www.forbes.com/sites/bernardmarr/2018/07/27/how-is-ai-used-in-healthcare-
5-powerful-real-world-examples-that-show-the-latest-advances/#610224b95dfb.
[Accessed 25 March 2021].
[29] B. Kalis, M. Collier and R. Fu. 10 pPromising AI Applications in Health
Care. Harvard Business Review 2018. Available at: https://hbr.org/2018/05/
10-promising-ai-applications-in-health-care. [Accessed January 10, 2021].
[30] S. Singhal and S. Carlton. The era of exponential Improvement in health-
care? McKinsey & Company; 2019.
[31] J.K. Kulski. Next-generation sequencing –an overview of the history, tools,
and "omic" applications. In: Next Generation Sequencing-Advances,
Applications and Challenges. London: Intech; 2016.
[32] J.P. Hughes, S. Rees, S.B. Kalindjian and K.L. Philpott. Principles of early
drug discovery. British Journal of Pharmacology. 2011; 162(6): 1239–1249.
[33] L. Zhang, J. Tan, D. Han and H. Zhu. From machine learning to deep
learning: progress in machine intelligence for rational drug discovery. Drug
Discovery Today. 2017; 22(11): 1680–1685.
[34] M. Rangaiah. Artificial Intelligence in Healthcare: Applications and Threats
| Analytics Steps; 2020. [online] Analyticssteps.com. Available at: https://
www.analyticssteps.com/blogs/artificial-intelligence-healthcare-applica-
tions-and-threats [Accessed 25 March 2021].
[35] A.N. Pisarchik, V.A. Maksimenko and A.E. Hramov. From novel technol-
ogy to novel applications: Comment on “An integrated brain-machine
interface platform with thousands of channels” by Elon Musk and Neuralink.
Journal of Medical Internet Research. 2019; 21(10): e16356.
[36] G. Litjens, T. Kooi, B.E. Bejnordi, et al. A survey on deep learning in
medical image analysis. Medical Image Analysis. 2017; 42, 60–88.
doi:10.1016/j.media.2017.07.005.
[37] Scnsoft.com. The State of the Art in Health Data Analytics; 2017. [online]
Available at: https://www.scnsoft.com/blog/health-data-analytics-overview
[Accessed 26 March 2021].
[38] V. Fortunati. Deep Learning Radiology: The Secret of Convolutional Neural
Networks; 2020. [online] Quantib.com. Available at: https://www.quantib.
com/blog/convolutional-neural-networks-in-deep-learning-radiology
[Accessed 25 March 2021].
[39] B. Kalis, M. Collier and R. Fu. 10 Promising AI Applications in Health
Care; 2018. Retrieved 5 April 2021, from https://hbr.org/2018/05/10-pro-
mising-ai-applications-in-health-care.
 Chapter 6
Blockchain in IoT healthcare: case study
Kripa Elsa Saji1, Nisha Aniyan1, Renisha P. Salim1,
Pramod Mathew Jacob1, Shyno Sara Sam1, Kiran Victor1
and Remya Prasannan1

Health sector is one of the most important industries in the world. With the recent
advent in the number of diseases, health is becoming the primary concern for each
individual. But with their busy schedules, they do not find enough time to act on
this concern. In this chapter, a health monitoring system has been proposed to sense
the vital body parameters and to store and share them in a secure environment. The
vital body parameters are sensed from the patient using sensors. The sensed data is
then stored into blockchain. The stored data can be accessed by the patient and
doctors using a web application. When a doctor needs to access the information
about a patient from another doctor, they only need to transfer the patient ID
between themselves to access the data.

6.1 Overview
Due to the recent advent in the number and kind of diseases, health has become the
primary concern for all humans. With the changing climate, sedentary lifestyle, and
other factors affecting the environment, diseases are finding a way to get in people’s
life stealthily. So, it is necessary to have a system that will be able to help people out
of this conflict. It is necessary to have a system that will help people out of this
predicament.
IoT or Internet of Things is a system of sensors and software used in the
collection and transfer data over a wireless network without any human commu-
nication. The sensors are used to collect data or information from an environment.
The sensed data are stored in a database via Wi-Fi, Bluetooth, Ethernet, etc. This
data is processed and depending on it, needed steps are taken.
Blockchain is a distributed ledger platform used to store transactions (blocks) in a
database (chain). The information stored in the blockchain is immutable. Blockchain

1
Department of Computer Science and Engineering, Providence College of Engineering, Chengannur,
APJ Abdul Kalam Technological University, India
112 The Internet of medical things

employs a peer-to-peer network that allows the transactions to imply at most two
parties (sender and receiver). It is a decentralized network so there is no chance to
interact with this directly through a third party. The data is securely stored crypto-
graphically. These technologies are getting implemented in many sectors for provid-
ing a secure environment, data protection, sensing data easily, etc. In this proposed
model, these technologies were incorporated to sense, monitor, store, and share health
parameters.
In the existing system, patients’ data is stored in electronic health record
(EHR) and electronic medical records systems. These systems are used to share the
medical information of the patient with health-care providers and organizations.
The various challenges faced by the user in this scenario are listed next:
● The users do not have access to their records stored in EHRs.
● When consulting another doctor, there is an overhead of paper-based medical
records.
● Paper-based records do not provide security, confidentiality, and integrity.
● The medical records in EHRs are not secure from third-party intrusion.
The proposed system addresses these challenges using IoT, blockchain, and a
web application. The model aims to provide secure storage along with monitoring
and sharing of vital body parameters.

6.2 Existing models to secure IoT healthcare

There are various EHR systems available based on various technologies. Most of
them focus on providing a secure and efficient way of medical information storage
using blockchain or cloud-based architectures. Here we discuss the summary of
some relevant systems available for securing IoT healthcare.
Kulkarni and Bakal [1] proposed a system to sense the body condition of
humans and then sent it to the medical experts. This medical team reviewed the
critical parameters that have been sensed and took further necessary action.
Kalaiselvi [2] proposed a system that is a network of connected Internet-
enabled devices that communicate with each other to take care of health using
remote sensors. This system does work without the assistance of human-to-human
or human-to-machine contact, this system can transmit data over a network. This
information is sent to the mobile applications.
Yeotkar and Gaikwad [3] proposed a model that is a remote monitoring tech-
nique which the person can be monitored wirelessly in his location using wearable
sensor arrangement by using accelerometer, temperature sensor, pulse oximeter and
heart-rate sensor, and galvanic skin response sensor to monitor physiological
human body.
Ved et al. [4] proposed a cloud-based personal health record system aiming for
an authorized user to access and review the patient’s records from any location. It is
a web application that works on the J2EE platform, deployed on Amazon EC2 and
runs on Microsoft SQL Server 2005. The user interface is divided into role-based
 Blockchain in IoT healthcare: case study 113

modules and for message passing and maintenance. The database layer includes the
DCM4CHE server and SQL server. For establishing security, password-protected
access is provided to all the users.
Joshi et al. [5] proposed a model for the health records to be accessed in a
delegated and secured manner using attribute-based encryption (ABE) for the man-
agement of access and data security in cloud-based EHRs. The EHR manager
application is a web application that is developed, using Python, based on the fun-
damentals of the model-view-controller architecture and is open-source. The system
consists of access broker (for authentication), EHR ontology (stores encrypted data),
ABE, and cloud service provider modules (for hosting EHR ontology).
Yang et al. [6] proposed a system to solve security problems and improve
storage efficiency. Cloud storage and blockchain technology are used to guarantee
the storage efficiency of electronic health records. ABE is used to assure the fine-
grained access control of the data in EHR which is part of security. The experi-
mental results showed the time for encryption and search to be independent to the
number of attributes.
Adlam and Haskins [7] proposed blockchain technological know-how has been
evolving and ought to perchance enhance the modern-day EHRs infrastructure. It is
suggested that EHRs often employ a role-based get entry to manipulate the model.
Permissioned blockchain technological know-how may want to enhance the
authorization model by influencing clever contracts and attribute-based get entry to
control. Hyperledger Fabric has been recognized as a permissioned blockchain
technological know-how ideal closer to use-cases that require privacy.
Mahore et al. [8] proposed a model based on cloud and blockchain which pro-
vides privacy and health-care data for statistical analysis to the researchers at the
same time. The model uses permissioned blockchain where the keys, certificates, and
all the participating entities are managed by the membership service provider. When
a patient visits a hospital, a cloud-based record is created and the metadata is stored
on the blockchain from which the researchers can obtain data. The patient has
complete control over his records, divided into sensitive and nonsensitive, and has
two key pairs for each. The sensitive data holding the information that can disclose
the patient’s identity is encrypted using public key, while the nonsensitive data that
holds the diagnosis data is encrypted using public key and stored on the cloud.
Ahmed et al. [9] proposed a distributed digital network system that acts as a
public ledger and uses bitcoin to operate money services, security services, IoT,
and various Internet applications. This system keeps the secure history of the
complete data for certain things. It has been explained as the shedding light of
blockchain, some of the typical algorithms used in the various blockchain, and
scalability issues.
Tanwar et al. [10] proposed a technology to strengthen virtualization by
implementing personalized healthcare in real-time using blockchain technology. The
blockchain can improve healthcare by using various solutions for eliminating the
prevailing limitations in health-care systems which includes tools and frameworks to
determine the effectiveness of the system such as Docker container, composer,
Hyperledger Fabric, Wireshark capture engine, and Hyperledger Caliper.
114 The Internet of medical things

Ramachandran et al. [11] proposed a blockchain technology that allowed the

balancing aspects of EHRs. Moreover, this technology assures the transactions
perform on EHRs are clear and restricted to make unauthorized changes. This
model aims to review the current blockchain-based pattern for the management of
EHRs and assess the scope of blockchain in the future.
Mary Subaja et al. [12] proposed blockchain technology know-how as a disbursed
method to provide safety in having access to the clinical record of a patient. It includes
three phases that are authentication, encryption, and data retrieval the usage of block-
chain technology. To withstand frequent attacks, quantum cryptography, Advanced
Encryption Standard, and secure hash algorithm algorithms are used for authentication,
encryption, and data retrieval, respectively. The proposed model guarantees the safety
of patient and trustworthiness and security of the health care system.
Liu et al. [13] proposed a scheme for the safety and sharing of clinical facts
based on the hospital’s blockchain to enhance the digital fitness device of the
hospital. The scheme can fulfill a variety of protection residences such as
decentralization, openness, and tamper resistance. A reliable mechanism is cre-
ated for the medical doctors to save clinical facts or get right of entry to the
historic facts of sufferers while meeting privateness preservation. Furthermore, a
symptom-matching mechanism is given between patients. Proxy re-encryption
technological know-how is used to assist the docs to access historical documents
of patients.
Radhakrishnan et al. [14] proposed a system that used a multilevel
authentication-based scheme to protect the blockchain from attacks and add one
more layer of security. If attackers change any data in the electronic health record,
it can prove to be fatal to the patients as it can result in wrong medication or
surgery. Blockchain plays a decisive role in securing data and transactions. The
challenge in this solution is scalability as the blockchain-based EHR system needs
large storage for storing data.
Nuetey Nortey et al. [15] proposed a system used for the preservation of
privacy during collecting, managing, and distributing EHR data. The system
ensures complete privacy, access control, and integrity of distributed EHRs to the
data owners during its dispersion on the blockchain. The proposed system inte-
grates blockchain system to achieve interoperability and data sharing with appro-
priate Application Programming Interfaces (APIs) with the use of its underlying
technology. Record validation in the system is done using modern cryptographic
techniques. The Hyperledger Fabric, a permissioned private blockchain having its
own characteristics and parameters, is used for implementation. But it provides the
facility for the creation of a private blockchain for research and testing purposes.
Huang et al. [16] proposed MedBloc which is a blockchain-based impervious
EHR device that permits patients and health-care carriers to get entry to and dis-
tribute fitness records in a privacy-preserving manner. MedBloc represents a
longitudinal view of the patient’s fitness history and permits patients to supply or
withdraw consent for managing rights of entry to their records. MedBloc is a tightly
closed EHR system that leverages blockchain technology. MedBloc protects
patients’ privacy by using the state-of-the-art encryption scheme.
 Blockchain in IoT healthcare: case study 115

Kim et al. [17] proposed a patient-centric medicinal drug history recording

gadget with the use of blockchain, which captures QR code printed on the envelope
with the aid of drug save based totally on prescription. The records are saved by
using hash fee of data in blockchain and prevent the altering of the data. This
machine used fast health-care interoperability resources, the worldwide fitness
records exchange standard, to enhance interoperability.
Harshini et al. [18] suggested a patient-driven model for record maintenance
and the usage of blockchain. The contracts can be integrated in future days
increasing its manageability in statistics exchange. This model deploys smart
contracts which executes when both the parties agree on a set of protocols. This
model suggests blockchain technology as one of the potential solutions for the
maintenance of health records in an efficient manner.
Mikula and Jacobsen et al. [19] proposed a system that aims to manage
identity and access to maintain authentication and authorization using blockchain.
The authentication and authorization of the users are done by the validation of
transactions in the blockchain. The proposed system includes a client application
(App), a database, an application server, and a server for authentication and
authorization. The communication of the user and system is done using the App
that, in turn, interacts with the application server to perform operations. The
blockchain stores the transactions as key-value pairs, where the key contains a
hash of the user’s public key.
Živi et al. [20] described tangle as DAG (directed acyclic graph) and its
advantages compared to other blockchain technologies. In tangle, each transaction
(child) points to two previous transactions (parents) and approves them. A transaction
can be approved multiple times in a direct or indirect way. The transactions become
immutable after reaching consensus, which is attained with the help of Random Walk
Monte Carlo algorithm. The tangle is consistent by using proof of work computation
that is performed by every transaction. The aggregate weight of a transaction is the
total number of successive transactions and the transactions that have affirmed it.

6.3 Blockchain to secure IoT healthcare

6.3.1 Proposed system
The proposed health monitoring system incorporates of the following three mod-
ules: sensor module, blockchain module, and the web application module. The
sensors are provided with the user to sense the required vital parameters. The data
sensed by the sensors will get stored into the blockchain. The web application will
permit the patient and doctors to access the data. The doctors can also consult
another doctor for a patient by sharing the patient ID rather than the whole data.
Figure 6.1 demonstrates the block diagram of the proposed system.

6.3.1.1 Sensor module

The proposed system aims to extract the vital body parameters from the user with
the help of sensors. The sensors used in the system are AD8232, DS18B20, and
116 The Internet of medical things

Retrieve the data

Sensed data IOTA
Sensors Patient

Retrieve the data

Consult
Consult
Doctor 2 Doctor 1

Figure 6.1 Architecture of using blockchain in IoT healthcare

MAX30100. The sensed parameters are read by NodeMCU. NodeMCU has inbuilt
Wi-Fi with the help of which this data will be sent to the blockchain.

6.3.1.2 Blockchain module

IOTA (name of a consortium) technology is a distributed ledger intended for the
IoT to hold up friction-less data and value transfer. The data from the server will be
stored in the IOTA. A seed is created, which is a private key used to login into the
IOTA wallet. Then generates an address and adds the sender address. And all
transactions are directed on the IOTA node. Tangle is used for authenticating the
transactions. All the data is securely stored in it. The users can access the patient
data stored in IOTA using the web application.

6.3.1.3 Web application module

The web application will permit the user to access and view their data that is stored
in IOTA. The user requests for the data stored in IOTA using IOTA tokens with the
help of client libraries. The data transactions are converted into trytes and then to
ASCII. This converted data will be provided to the user. The users can also share
their data with other participating entities. When users share their data with the
doctors, the doctors can update the data by adding their suggestions.

6.3.1.4 Working of proposed system

The working of the proposed system is illustrated in Figure 6.2. Sensors are used to
sense the vital body parameters. The sensed data will be read using NodeMCU and
stored into the blockchain (IOTA). The users can access, view, and share their data
 Blockchain in IoT healthcare: case study 117

Patient
Blockchain
Node MCU

Send data Retrieve data

Sensed data

Retr
ata

ieve
ed
riev

data
Ret

Consult
Sensors

Consult

Doctor 2 Doctor 1

Figure 6.2 Working of proposed blockchain system for IoT healthcare

MAX30100

IOTA
DS18B20 NodeMCU

AD8232

Figure 6.3 Block diagram of sensor layer

with other entities using the web application. The doctors can also consult another
doctor by only sharing the patient ID rather than the whole medical record.
The block diagram of sensors layer is shown in Figure 6.3. The sensors used in
the system for sensing the vital body parameters are MAX30100 (for pulse and
saturation), DS18B20 (for temperature), and AD8232 (for ECG). The micro-
controller unit NodeMCU is used to read the data that is sensed using the sensors.
The data is then stored into the blockchain platform, IOTA.
In Figure 6.4, the use case diagram demonstrates the various use cases of the
health monitoring system. The end users of the system will be users and doctors.
The user can view the sensed data, share the data with other users (doctors), and
view their updated data through the web application. The doctor can modify the
118 The Internet of medical things

Health monitoring system

Store the sensed data

Retrieval of stored data

Consult a doctor

Patient Share the patient ID Doctor

Update the patient information

Figure 6.4 Use case diagram of health monitoring system

medical records of the patient. The doctor can also refer another doctor by sharing
the patient ID, used for retrieving the patient’s medical records.

6.3.2 System implementation

The system is implemented using multiple technologies both on the software front
and hardware sides. On the hardware side, sensors are consolidated with the
microcontroller NodeMCU and the detected data is stored into the blockchain. On
the software side, a web app is developed for retrieval and sharing of the
stored data.
Multiple sensory data can be read from a patient which can be useful in
determining their health conditions or even some underlying health problems.
These data points are read using multiple sensors that are integrated into the sys-
tem. The data is then stored into IOTA—a blockchain solution—to ensure the
utmost security of the patient data. The web application is then used to access this
data on request. The various hardware components required are
● pulse oximeter and heart rate sensor
● ECG sensor
● temperature sensor
● NodeMCU

6.3.2.1 Pulse oximeter and heart rate sensor

The sensor detects the pulse rate and level of oxygen saturation of the body. It uses
two LEDs (emits infrared and red lights), optimized optics, photodetector, and low-
noise analog signals. For detecting pulse rate, only the infrared LED is used and for
oxygen saturation level, both red and infrared LEDs are used. The libraries
MAX30100_PulseOximeter and Wire are used to read the data from MAX30100
 Blockchain in IoT healthcare: case study 119

and to communicate with MAX30100 using I2C protocol, respectively. The pulse
oximeter and heart rate sensor are shown in Figure 6.5 (left).

6.3.2.2 ECG sensor

The sensor is used to determine the electrical movement of the human heart. It
amplifies, extracts, and filters the biopotential signals in noisy conditions. It
includes pins to connect the ECG electrodes and an LED to indicate the rhythm of
heartbeat of a human. Figure 6.5 (right) shows the ECG sensor (AD8232).

6.3.2.3 Temperature sensor

The sensor detects the body temperature of the human body. It is a 1-wire interface
sensor that requires only one digital pin to communicate to and from the micro-
controller. It includes a unique 64-bit serial code that allows multiple DS18B20s to
operate on the same 1-wire bus, making it easy for one microcontroller to control
multiple distributed DS18B20s. The libraries OneWire and Dallas Temperature are
used to communicate with DS18B20 and to access the registers of DS18B20 for
reading temperature, respectively. Temperature sensor (DS18B20) is shown in
Figure 6.6 (left).

MAX30100 sensor
AD8232 sensor

Figure 6.5 Pulse oximeter and heart rate sensor (left) and ECG sensor (right)

Temperature sensor
NodeMCU

Figure 6.6 Temperature sensor (left) and NodeMCU board (right)

120 The Internet of medical things

6.3.2.4 NodeMCU
NodeMCU is an open-source IoT firmware and development kit that is designed to
develop IoT-based applications. It consists of multiple digital and one analog pin. It
also has an inbuilt Wi-Fi that enables to communicate the detected data with the
Internet. NodeMCU is shown in Figure 6.6 (right).
The circuit diagram of the system is represented in Figure 6.7. All the sensors are
integrated together to the central coordinator which is NodeMCU in this scenario.
Hardware connections of the sensor module are given next:
1. Connect VIN pin of MAX30100 to the 3.3V pin of NodeMCU.
2. Connect INT pin of MAX30100 to the D0 pin of NodeMCU.
3. Connect SCL pin of MAX30100 to the D1 pin of NodeMCU.
4. Connect SDA pin of MAX30100 to the D2 pin of NodeMCU.
5. Connect GND pin of MAX30100 to the GND pin of NodeMCU.
6. Connect 3.3V pin of AD8232 to the 3v3 pin of NodeMCU.
7. Connect GND pin of AD8232 to the GND pin of NodeMCU.
8. Connect OUTPUT pin of AD8232 to the A0 pin of NodeMCU.
9. Connect LO
pin of AD8232 to the D5 pin of NodeMCU.
10. Connect LOþ pin of AD8232 to the D6 pin of NodeMCU.
11. Connect the left pin of DS18B20 to the GND pin of NodeMCU.
12. Connect the middle pin of DS18B20 to the 3v3 pin of NodeMCU.
13. Connect the right pin of DS18B20 to the D4 pin of NodeMCU.

Figure 6.7 Circuit diagram of the system

 Blockchain in IoT healthcare: case study 121

6.3.2.5 IOTA
IOTA is an open, scalable distributed ledger that allows feeless transactions
designed for an IoT ecosystem. It is based on tangle, a DAG. In tangle, every
transaction points to two preceding transactions that are called as parents and get
validated by the child transaction. Tangle grows in the direction of the paths that
have the highest weights and the lightest branches never reach consensus.
Figure 6.8 depicts the transactions in a tangle.
In IOTA, all the nodes issue and validate transactions, i.e., a node validates
previous two transactions and then issues its own transaction. IOTA is asynchro-
nous. A transaction takes different amount of time to reach different nodes and
another node may be seeing a previous version of the tangle. The nodes always
share new transactions with each other to increase their probability of getting
verified and eventually reaching the consensus.

6.3.2.6 Masked-authenticated messaging

Masked-authenticated messaging (MAM) is an IOTA protocol-based module that
permits sending and reading message flows through tangle. It allows to publish the
sensor data on the IOTA tangle. Cryptography and authentication are applied to
assure that the message will not be doctored and comes from an authenticated
sender. MAM uses a Merkle signature scheme that provides more secure environ-
ment by resisting quantum computer attacks and short generating and verifying
time period for signatures.
MAM has three modes of operation: public, private, and restricted. In public
mode, root of the Merkle tree is taken as the address of a transaction. In private
mode, the hash of the root is the address and in restricted mode, the address has to

Tip New tip

Unconfirmed Conflicting
w
Fully confirmed Validation path q

s
m x

j u
g p
b
z
c k n v
i
a
d f
y
o t
e l
h
r
D

Figure 6.8 Transactions in tangle

122 The Internet of medical things

be calculated from the root and search for masked messages (obtained using root
and a sideKey). Figure 6.9 depicts the data stored in the public mode of MAM.

6.3.2.7 MongoDB
MongoDB is a document-oriented NoSQL database that is suitable for unstructured
data. The data store is lightweight, easy to traverse, and allows fast execution.
MongoDB Atlas is used to store the MongoDB database that is being used with the
system. MongoDB Atlas is a cloud provider that lets users host their databases and
related APIs. Integrating it into the system allows it to scale exponentially and also
increases the availability of the system.

6.3.2.8 NodeJS
NodeJS is an open-source server platform based on JavaScript. It facilitates the
creation of scalable and fast network applications. It provides a runtime environ-
ment and various JavaScript modules for the development of web applications.
NodeJS is asynchronous (non-blocking) resulting in scalable and fast execution
of code.

6.3.2.9 Express.js
Express.js is a web application framework based on NodeJS which provides robust
APIs to build web apps and websites. Many modules are available in npm which
can be plugged into Express making it flexible and pluggable. It helps to manage
server and the various routes associated with it.

6.3.2.10 Storing the sensed data in IOTA

The health parameters are read using the sensors which is then fed into the
NodeMCU. The NodeMCU sends the data to a NodeJS server which then processes

1st generation
0-th generation
(genesis) To next generation
<title>My secret
2</title>
<title>My secret 1</title> <div>Furthermore,
nextRoot nextRoot
<div>To tell the truth... ...

Decrypt

Given info.
Decrypt

Root Encrypted trytes Encrypted trytes

Address Address

Figure 6.9 Public mode of MAM

 Blockchain in IoT healthcare: case study 123

the data to work with the other components in the workflow. The data is sent to the
NodeJS server because the NodeMCU does not have enough power to directly
write the data into the IOTA.
Express.js is used at the NodeJS server to further enhance the networking
capabilities of the NodeJS server and also enable routing for the server. Express
allows the processing of the incoming of the POST requests and also helps with the
generation of the response and redirection of the data to other parts of the archi-
tecture. The extracted data is correlated with the base database hosted on the
MongoDB Atlas, and if a match is found, the sensed data is written into the IOTA
Devnet using the MAM package.
When the patient data is published for the first time, the root node will not be
present in the MongoDB. In this case, the root is stored in the MongoDB and after
storing the data, the latest state will be stored into a JSON file. The latest state will
be an object that consists of the seed and the configurations of the state. The latest
state is used to connect the transactions in the tangle. This means that all the data of
a patient will be connected using the latest state. Figure 6.10 shows the data blocks
stored in IOTA tangle using MAM.

6.3.2.11 Retrieving the data using web application

The process of retrieval of patient data is as follows. When patient ID is entered, all
matching documents from MongoDB are retrieved. When the root file is fed into
the mam.fetch() function, a set of callback functions is retrieved. The first callback
function retrieves the data of each block and the second callback functions print all
the data in the tangle. Each block of data holds the hash to access the next block.
This hash can be given to the first callback function to retrieve the next block.
Using this technique, all blocks are retrieved and then stored into an array.
When a block is received, the hash of the next block is taken and fed into the

Figure 6.10 Data stored in IOTA using MAM

124 The Internet of medical things

Figure 6.11 Fetching data from IOTA network using MAM

callback function that was received. This process is done until the last block. Then
the second callback function is used to display all this data. These data trytes are then
converted to ASCII and then to JSON. Express.js is used to render this JSON data as
a table. A user who is identified as a patient can only retrieve their own data. When a
patient tries to retrieve the data by providing a patient ID, the requested ID and the
patient ID are verified. If a match is found, the data is retrieved. Since this application
is ran on development parameters, the public mode of MAM is used for better sta-
bility and predictability. Figure 6.11 represents fetching data from MAM explorer.

6.4 Conclusions
The proposed health monitoring and management system consists of a hardware
module, blockchain module, and web application. The sensors incorporated with
the NodeMCU detect the health data from a patient’s body which can be used for
determining their health conditions. The data is stored into an IOTA node with the
help of a server and MAM. The patient ID is used to retrieve the stored data using
the Web application. A patient can only retrieve his/her own data. After retrieving
the data, the patient can consult with the doctor or a doctor can refer another doctor.
When a doctor consults another doctor, the patient ID is shared using which the
consulted doctor can access the patient’s data. The advantages of the proposed
system are as follows:
● Easier to track and analyze the vital parameters.
● Users can access their information and can use it to consult with a doctor.
● The overhead of paper-based records is eliminated.
● The process of consulting another doctor is made easier.
● More economical and optimum utilization of resources.
 Blockchain in IoT healthcare: case study 125

The future plans are to incorporate a device with more processing power in the
hardware to eliminate the use of server and publish the data directly to IOTA
tangle. Our next stage will try to publish the data blocks in the restricted mode of
MAM and provide access control on the data stored on IOTA tangle.

References

[1] N. J. Kulkarni and J. W. Bakal, “Real Time Vital Body Parameter Monitoring,”
in Fourth International Conference on Computing Communication Control
and Automation (ICCUBEA), 2018.
[2] G. Kalaiselvi, “A Comprehensive Study On Healthcare Applications using
IoT, “International Journal of Engineering Science Invention (IJESI),
pp. 41–45, 2018.
[3] H. S. Yeotkar and V. T. Gaikwad, “IoT Based Human Body Parameters
Monitoring by Using Wearable Wireless Sensor Network,” International
Research Journal of Engineering and Technology (IRJET), vol. 06, no. 07,
pp. 2458–2466, 2019.
[4] V. Ved, V. Tyagi, A. Agarwal, and A. S. Pandya, “Personal Health Record
System and Integration Techniques With Various Electronic Medical Record
Systems,” in IEEE 13th International Symposium on High-Assurance
Systems Engineering, Boca Raton, 2011.
[5] M. Joshi, K. Joshi, and T. Finin, “Attribute Based Encryption for Secure
Access to Cloud Based EHR Systems,” in IEEE 11th International
Conference on Cloud Computing, 2018.
[6] X. Yang, T. Li, R. Liu, and M. Wang, “Blockchain-Based Secure and
Searchable EHR Sharing Scheme,” in 2019 4th International Conference on
Mechanical, Control and Computer Engineering (ICMCCE), Hohhot, China,
2019.
[7] R. Adlam and B. Haskins, “A Permissioned Blockchain Approach to the
Authorization Process in Electronic Health Records,” in 2019 International
Multidisciplinary Information Technology and Engineering Conference
(IMITEC), Vanderbijlpark, South Africa, 2019.
[8] V. Mahore, P. Aggarwal, N. Andola, Raghav, and S. Venkatesan, “Secure
and Privacy Focused Electronic Health Record Management System using
Permissioned Blockchain,” in 2019 IEEE Conference on Information and
Communication Technology, Allahabad, India, 2019.
[9] I. Ahmed, Shilpi, and M. Amjad, “Blockchain Technology A Literature
Survey,” International Research Journal of Engineering and Technology
(IRJET), vol. 5, no. 10, 2018.
[10] S. Tanwara, K. Parekha, and R. Evans, “Blockchain-Based Electronic
Healthcare Record System for Healthcare Applications,” Journal of
Information Security and Applications, 2019.
[11] S. Ramachandran, O. Obu Kiruthika, A. Ramasamy, R. Vanaja, and S.
Mukherjee, “A Review on Blockchain-Based Strategies for Management of
126 The Internet of medical things

Electronic Health Records (EHRs),” in 2020 International Conference on

Smart Electronics and Communication (ICOSEC), Trichy, India, 2020.
[12] C. Mary Subaja, M. A. Anigo, S. G. Partha, C. Priyanka, and M. Raj Kumari,
“An Efficient Data Security in Medical Report using Block Chain
Technology,” in 2019 International Conference on Communication and
Signal Processing (ICCSP), Chennai, India, 2019.
[13] X. Liu, Z. Wang, C. Jin, F. Li, and G. Li, “A Blockchain-Based Medical
Data Sharing and Protection Scheme,” IEEE Access, vol. 7, pp. 118943–
118953, 2019.
[14] B. L. Radhakrishnan, A. Sam Joseph, and S. Sudhakar, “Securing
Blockchain based Electronic Health Record using Multilevel
Authentication,” in 2019 5th International Conference on Advanced
Computing & Communication Systems (ICACCS), Coimbatore, India, 2019.
[15] R. Nuetey Nortey, L. Yue, P. Ricardo Agdedanu, and M. Adjeisah, “Privacy
Module for Distributed Electronic Health Records (EHRs) Using the
Blockchain,” in 2019 IEEE 4th International Conference on Big Data
Analytics (ICBDA), Suzhou, China, 2019.
[16] J. Huang, Y. Wei Qi, M. Rizwan Asghar, A. Meads, and Y-C. Tu,
“MedBloc: A Blockchain-Based Secure EHR System for Sharing and
Accessing Medical Data,” in 2019 18th IEEE International Conference On
Trust, Security And Privacy In Computing And Communications/13th IEEE
International Conference On Big Data Science And Engineering (TrustCom/
BigDataSE), Rotorua, New Zealand, 2019.
[17] J. W. Kim, A. R. Lee, M. G. Kim, I. K. Kim, and E. J. Lee, “Patient-Centric
Medication History Recording System Using Blockchain,” in 2019 IEEE
International Conference on Bioinformatics and Biomedicine (BIBM), San
Diego, CA, USA, 2019.
[18] V. M. Harshini, S. Danai, H. R. Usha, and M. R. Kounte, “Health Record
Management through Blockchain Technology,” in 2019 3rd International
Conference on Trends in Electronics and Informatics (ICOEI), Tirunelveli,
India, 2019.
[19] T. Mikula and R. H. Jacobsen, “Identity and Access Management with
Blockchain in Electronic Healthcare Records,” in 2018 21st Euromicro
Conference on Digital System Design (DSD), Prague, 2018.
[20] N. Živi, E. Kadušić, and K. Kadušić, “Directed Acyclic Graph as Tangle: An
IoT Alternative to Blockchains,” in 2019 27th Telecommunications Forum
(TELFOR), Belgrade, Serbia, 2019.
 Chapter 7
Adaptive dictionary-based fusion of
multi-modal images for health care
applications
Aishwarya Nagasubramanian1, Chandrasekaran
Bennila Thangammal2 and Vasantha Pragasam
Gladis Pushparathi3

7.1 Introduction
Medical imaging has made significant advances in a variety of health-care uses in
recent years, including research, diagnosis, and education. Various imaging mod-
alities such as computed tomography (CT), magnetic resonance imaging (T1-MRI
and T2-MRI), positron emission tomography (PET), and single-photon emission-
computed tomography (SPECT) are used to provide more clinical data to medical
practitioners, which reflects different information about the human body [1,2].
However, due to technical limitations, a single modality image cannot give enough
information to meet specialists’ clinical needs. Combining information from different
modalities (CT-MRI, PET-MRI, SPECT-MRI, etc.) into a single image, which better
identifies the region of interest and enhances diagnostic accuracy, is an effective way
to overcome this problem. Despite the fact that there are numerous algorithms to
conduct this operation [3,4], dictionary-based sparse representation (SR) techniques
have grown in popularity in the field of image fusion. The first is a fixed dictionary
that has been built using analytical models such as the discrete cosine transform
(DCT), wavelet, etc. The second is a learned dictionary made from high-quality
natural images. In image fusion, Yang and Li [5] were the first to introduce sparse
representation. For multi-focus image fusion, they used a redundant DCT dictionary.
Aharon et al. [6], on the other hand, demonstrated that an adaptive dictionary learned
from input visuals beats a fixed dictionary. As a result, Yu et al. [7] suggest a fusion

1
Department of Electronics and Communication Engineering, Amrita School of Engineering, Amrita
Vishwa Vidyapeetham, Chennai, India
2
Department of Electronics and Communication Engineering, R.M.D. Engineering College
Kavaraipettai, Chennai, India
3
Department of Computer Science and Engineering, Velammal Institute of Technology Panchetti,
Chennai, India
128 The Internet of medical things

strategy based on a joint sparsity model. The dictionary is trained by simply adding
all of the source image patches, which takes a long time and renders the dictionary
ineffective. For simultaneous fusion and denoising of multimodal pictures, Yin and
Li [8] used a joint sparsity model. The experimental findings were adequate, although
the proposed method’s performance is significantly reliant on external data. For
sparse coefficient estimation, Yang and Li [9] introduced the simultaneous ortho-
gonal matching pursuit algorithm (OMP). The proposed method is tested on a variety
of image types, and the dictionary used in this method is very sensitive on the input
image. Wang et al. [10] combined non-subsampled contourlet transform (NSCT)
with SR to perform image fusion. A major benefit of this strategy is that it produces
comparable experimental findings with less computing time.
From the focus information map of multi-focus images, Nejati et al. [11]
suggested a global dictionary learning approach. The dictionary input training data
is created by randomly selecting an arbitrary number of patches, which may or may
not result in a highly organized dictionary. Liu et al. [12] used multi-scale trans-
forms (MSTs) and SR to offer an image fusion framework. For the dictionary
learning process, a collection of 40 high-quality natural pictures is used. The same
author later suggested an adaptive sparse representation (ASR) model for image
fusion and denoising at the same time. The histogram of gradients is used to learn
six compact sub-dictionaries. The disadvantage of this approach is that it has a
higher computational cost. Kim et al. [13] suggested a dictionary learning strategy
based on K-means clustering and principal component analysis to lower the com-
putational cost of SR-based fusion approaches. The learnt dictionary is compact
and informative, but with this method, the number of clusters must be preset.
As a result, existing SR-based fusion solutions either share a preconstructed
vocabulary that requires previous knowledge of the external pre-collected data or uti-
lize a learnt dictionary that requires prior knowledge of the external pre-collected data.
The global dictionaries provided by Aishwarya and Bennila Thangammal [14] have
been shown to have greater representation ability than existing SR-based approaches;
nonetheless, there are still several key difficulties that need to be addressed. To begin,
previous knowledge of the external data sets is required for the creation of the initial
training samples. In reality, collecting appropriate training samples may not be
achievable. Furthermore, the fusion performance is highly dependent on the original
training data set’s construction. Second, global dictionaries are built using the inherent
structures of high-resolution images. As a result, dictionaries of this sort are more
suited to single-sensor image fusion, such as multi-focus image fusion. Learning a
dictionary adaptive to the relevant source images, on the other hand, can be more
efficient in representing the complex, intrinsic structures of multi-sensor or multimodal
fusion applications. As a result of these techniques, the need for a discriminative over-
complete dictionary that reflects the complicated structures of given images has arisen.
The design of the dictionary, activity-level measurement of sparse coefficients,
and the design of the optimum fusion rule are the significant elements that deter-
mine the ultimate fusion quality of SR-based techniques. Most SR-based fusion
algorithms have focused on the former element so far, while the latter part has not
been adequately addressed. Due to the aforementioned difficulties, this chapter
 Adaptive dictionary-based fusion 129

proposes a multimodal medical fusion method based on adaptive dictionary

learning. Because the adaptive dictionary is trained based on the modified spatial
frequency (MSF) indicator and used for the fusion of medical images, the suggested
method is called MIFMSF.

7.2 Learning a dictionary

The main goal of this work is to create a comprehensive and compact over-
complete dictionary in order to improve fusion performance. An informative
sampling strategy based on MSF is proposed to generate such a dictionary.
Figure 7.1 depicts the detailed dictionary learning process. Assume the source
images IA and IB are registered with size N
N. To begin, each source image
IA and IB is divided into image patches of size
n
n using the sliding window
T
approach. Let the patches be denoted by P ¼ ptc t¼1 ; c 2 fA; Bg. Generally, only
a significant percentage of medical image patches contain clinically valuable
information. Using all of the image patches from the source images makes the
dictionary very redundant and has a significant impact on the fusion result. The
computing time is also increased as a result of this learning process. In order to
address the aforementioned difficulties, only informative patches with higher
structural information are chosen for dictionary learning.
First, any patch ptc with an intensity variance of less than five ðvarðptc Þ < 5Þ
will be eliminated. This preprocessing step removes all the meaningless patches of
source images. The source image patches obtained after the preprocessing phase

N1
N2
are denoted by PA ¼ p jA j¼1 and PB ¼ p kB k¼1 . Training a dictionary with
informative patches is widely known for allowing richer data representation than
typical preconstructed dictionaries. The structural information of patches is

Informative sampling

Select/ignore the patch

Overlapping Preprocessing of Evaluation of structural based on selection criterion
Input images
image patches image patches in content using focus based on threshold
set P using measure
variance

Construct
training data set

K-SVD

Adaptive dictionary

Figure 7.1 Overall framework of proposed dictionary learning approach

130 The Internet of medical things

weighed using MSF to select such informative patches. MSF [15] is a useful metric
for determining an image’s overall edge strength in gradient directions. The MSF of
an image I with size M1
M2 is calculated as follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
MSF ¼ RF 2 þ CF 2 þ SDF 2 þ MDF 2 (7.1)
where RF and CF are the row and the column frequency which is given by
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 1 X M1 X M2
RF ¼ t ½Iðx; yÞ  Iðx; y  1Þ2 (7.2)
M1 M2 x¼1 y¼2
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 1 X M1 X M2
CF ¼ t ½Iðx; yÞ  Iðx  1; yÞ2 (7.3)
M1 M2 x¼2 y¼1

and frequencies along the main diagonal and secondary diagonal are given by
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 1 X M1 M X 2 1

SDF ¼ t ½Iðx; yÞ  Iðx  1; y þ 1Þ2 (7.4)

M1 M2 x¼2 y¼1
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u 1 X M1 X M2
MDF ¼ t ½Iðx; yÞ  Iðx  1; y  1Þ2 (7.5)
M1 M2 x¼2 y¼2

Equation (7.1) is used to calculate

the
Nedge strength of each

patch
Nin PA and PB ,
which is indicated as MSFA ¼ MSFAj j¼1 and MSFB ¼ MSFBk k¼1 : In (7.6), a
1 2

cutoff threshold Tc ; c 2 A; B, is chosen to improve the efficiency of the dictionary

learning process:
Tc ¼ 0:1
maxðMSFc Þ; c 2 A; B (7.6)

The threshold value is set low enough that both low- and high-frequency pat-
ches are selected for training with minimal redundancy. For the dictionary learning
process, let TD; fTDl jl ¼ 1; 2; 3; :::; Lg indicate the training data set. To deter-
mine whether the patch contains enough spatial features, a selection rule is used.
This is accomplished by comparing MSFA and MSFB of each patch in PA and PB
with Tc . For example, if MSFAj > TA ; select p jA . Otherwise ignore p jA :

Select p jA ; if MSFAj > TA
TDl ¼ (7.7)
Ignore p jA ; otherwise:
According to the previous criteria, the patch with fine details and superior
visual clarity is used in the dictionary learning process. This procedure is repeated
until all of the useful patches in PA and PB have been retrieved. The mean value of
each informative patch in the training data set is subtracted from zero before the
dictionary learning process to guarantee that only the edge structures of the patches
 Adaptive dictionary-based fusion 131

are included for training. Finally, the K-SVD method iterates between the sparse
coding step and the dictionary update step D to learn an adaptive dictionary D.

7.3 Experimental setup and analysis

The size of the image patch used to create the training data is set to 8
8. The
K-SVD algorithm is used to learn an adaptive dictionary with a size of 400 words
and 50 iterations. Let us say each source image is having the size of 256
256.
There will be 124,002 input patches for dictionary learning if the patch size is 8
8
and the step length is set to 1 pixel. According to the proposed method, the training
data set TD contains about 40,000 patches, or about 32% of the total source image
patches. Obtaining an adaptable dictionary takes around 1,500 s. As a result, pro-
cessing time is reduced and memory utilization is optimized. Furthermore, the
dictionary’s redundant information is eliminated to a greater extent.

7.4 Overview of fusion scheme

Figure 7.2 shows a schematic diagram of the MIFMSF fusion scheme. The basic
flow is comparable to standard SR-based fusion approaches, with the exception of
sparse coefficient fusion. The input medical images are separated into overlapping
image patches and lexicographically ordered into column vectors denoted by

Sparse representation
Coefficient fusion and reconstruction

Normalized Sparse Fused

image estimation sparse
vectors Activity-level coefficients
IA using OMP measurement
using MSF

Fused
vectors

IB

Fused image

Figure 7.2 Overview of MIFMSF fusion scheme

132 The Internet of medical things

t T
T
VA ¼ vA t¼1 and VB ¼ vtB t¼1 . Before sparse coding, the DC component (mean
value) of each patch is normalized to zero.
For sparse coefficient estimation, the batch-OMP technique is used to improve
computing efficiency. In the fusion of sparse coefficients, there are two major
challenges to consider. The first issue focuses on determining the activity level of
source images, which aids in identifying their different features. The second issue
focuses on the integration of sparse coefficients into fused image counterparts.
The former challenge is overcome in traditional SR-based fusion algorithms by
using the L1-norm, which describes the detail information contained in the sparse
vectors. The higher the value, the more energy the related image patch has.
However, this metric ignores the image patch’s spatial details, which are critical for
preserving the source images’ original visual clarity. Because SR is an approx-
imation approach, measuring activity levels with a single measurement may not be
possible. Due to the aforementioned issue, it has been suggested that the activity
level of picture patches be measured in both the spatial and transform domains.
MSF, as mentioned in the preceding section, describes the spatial features that an
image contains. The higher the value, the more information it contains. The
information from L1-norm and MSF is combined for each source image patch
using scalar multiplication. The sparse coefficients of the tth patch in IA and IB are
given by atA and atB . The notion kk1 denotes the L1-norm of sparse vectors and
MSFAt and MSFBt are the corresponding MSFs. The proposed activity level mea-
surement is calculated by

wtA ¼ katA k1
MSFAt (7.8)

wtB ¼ katB k1
MSFBt (7.9)
where wtA and wtB are the measurement results for the tth patch of source images
IA and IB .
Existing SR-based fusion approaches solve the latter issue by employing either
a weighted average fusion rule or a maximum absolute (max-abs) rule. The first
rule combines the input sparse vectors with certain user-defined weights based on
the activity level to produce fused coefficients. This rule has the propensity to
reduce the contrast of the fused image while suppressing the original texture of the
input images. The other criterion chooses the sparse vector with the highest abso-
lute value, allowing the fused image to capture the most important aspects of the
source images. In general, a fusion rule should be designed to absorb all of the
significant visual information from the input images into the fused image. In this
chapter, the maximum weighted activity level is chosen as the fusion rule based on
the aforementioned parameters. The proposed fusion rule is defined as follows:

atA ; if wtA > wtB
atF ¼ (7.10)
atB ; otherwise

The fusion coefficients are formed by selecting the sparse vector having
maximum activity level both in spatial domain and transform domain. The fused
 Adaptive dictionary-based fusion 133

vector for the tth patch is obtained as

vtF ¼ DatF þ mtF :1 (7.11)

where the fused mean value of tth vector mtF is given by
 t
mA ; if atF ¼ atA
mF ¼
t
(7.12)
mtB ; otherwise

7.5 Simulation results and discussion

Figure 7.3 shows five standard multimodal medical data sets that are used to
evaluate the performance of the proposed MIFMSF fusion scheme. These data sets
are divided into two categories. CT and MRI picture pairs make up the first group
(image data sets 1, 2, and 3), while T1-weighted MRI (MR-T1) and T2-weighted
MRI (MR-T2) image pairs make up the second group (image data sets 4 and 5). The
photos are all the same size.

7.5.1 Results on standard multimodal medical data sets

Some recent SR-based methods, such as MST-based sparse representation (MST-
SR) ASR model, and other state-of-the-art methods, such as NSCT proposed by
Zhang and Guo [16], multilevel local extrema proposed by Hu [17], and guided
filtering (GFF) based on guided filtering proposed by Li et al. [2], are compared to
show the potency of the MIFMSF method.
Figure 7.4 shows the fusion findings of a CT and MRI image pair for subjective
evaluation. The bone structure is visible in the CT picture without any loss of
information. An MRI scan depicts the internal organs and soft tissues in great detail.
The MIFMSF approach, as shown in Figure 7.4, retains the complimentary
information of source pictures with the best esthetic appearance and the fewest

1 2 3

4 5

Figure 7.3 Medical image data sets: 1–3—CT and MRI images; 4 and 5—MR-T1
and MR-T2 images
134 The Internet of medical things

artifacts. The NSCT method’s fusion result in Figure 7.4(c) loses a lot of image
features and has a contrast reduction problem. Figure 7.4(d) shows the blocking
artifacts in the fusion outcome of multilevel local extrema. The outputs of GFF and
MST-SR fusion are more visible; however, they do not maintain certain local
information from the original images. Figure 7.4 shows that the fusion results of the
ASR method and the MIFMSF approach cannot be separated easily. As illustrated
in Figure 7.5, a particular portion of source pictures is expanded. Figure 7.5(c)
shows that the NSCT approach has lost the edge and texture information of the
original images, resulting in low contrast. The multilevel local extrema method has

(a) (b) (c) (d) (e)

(f) (g) (h) (i)

Figure 7.4 Fusion results of CT and MRI image pair: (a) and (b) source images;
fusion results of (c) NSCT, (d) multilevel local extrema, (e) GFF, (f)
MST-SR, (g) ASR-128, (h) ASR-256, and (i) MIFMSF method

(a) (b) (c) (d) (e)

(f) (g) (h) (i)

Figure 7.5 The magnified details of fused image: (a) and (b) regions of source
images. Fusion details of (c) NSCT, (d) multilevel local extrema, (e)
GFF, (f) MST-SR, (g) ASR-128, (h) ASR-256, and (i) proposed method
 Adaptive dictionary-based fusion 135

Table 7.1 Average statistical evaluation of five standard medical image pairs for
different methodologies

Methods QAB/F QCB QMI Qw Q Qe MG

NSCT 0.4547 0.5312 0.6965 0.4753 0.8098 0.3976 7.0088

Multilevel local extrema 0.602 0.566 0.9018 0.7427 0.8597 0.5669 9.7644
GFF 0.6509 0.5886 0.7177 0.6871 0.8489 0.5465 8.497
MST-SR 0.6382 0.6085 0.7426 0.7749 0.8491 0.621 9.673
ASR-128 0.5849 0.583 0.6701 0.6426 0.8196 0.5312 7.8524
ASR-256 0.5619 0.577 0.6673 0.6151 0.8158 0.4947 7.628
MIFMSF method 0.6606 0.6129 0.9056 0.7755 0.8762 0.6299 9.1806

substantial artifacts that blur the fused image’s edge region (Figure 7.5(d)).
Artificial traces (see the eye regions in Figure 7.5(e)–(h)) are introduced in the
fused image since the GFF, MST-SR, and ASR algorithms fail to capture the ori-
ginal information of source images.
When compared to the input images, the MIFMSF technique produces output
with greater sharpness, clarity, and information content. Furthermore, the proposed
MIFMSF approach is free of undesired degradations and blocking artifacts, making
it ideal for precise clinical diagnosis.

7.5.2 Objective analysis of standard medical image pairs

The average quantitative comparison of five standard multimodal medical image
pairs are compared and reported in Table 7.1. The best results are indicated in bold.
Table 7.1 shows that, with the exception of mean gradient (MG), the suggested
technique uses all of the highest quantitative results. The largest value of MG is
found in the multilevel local extrema method, which could be attributed to unde-
sirable artifacts in the fused image. Furthermore, the metric MG is only evaluated
in terms of the final fused image. As a result, this statistic alone cannot determine
the proposed method’s fusion performance. The MIFMSF approach focuses on
preserving source picture edge features and information content with maximum
visual clarity. Higher value of QAB=F ; QCB ; QMI ; Q; Qw and Qe demonstrates the
effectiveness of proposed MIFMSF method to a greater extent.

7.6 Summary
The flaws of existing SR-based fusion schemes are investigated, and a dis-
criminative dictionary with less computational effort is constructed without any
prior knowledge about the external data sets. From the experimental analyses, the
discriminative dictionary is constructed with approximately 32% of the total source
image patches. The new fusion rule modifications ensure that all of the critical
spatial information of the source images is well retained in the final fused image,
resulting in the best visual appearance. As compared to state-of-the-art SR-based
136 The Internet of medical things

fusion schemes, the MIFMSF method shows approximately 4% improvement in

quantitative analysis under noise-free condition. It has been experimentally verified
that the MIFMSF technique shows promised results and competes with most of the
mainstream techniques both visually and quantitatively.

References

[1] V. Barra and J. Y. Boire, A general framework for the fusion of anatomical
and functional medical images, NeuroImage 13 (2001), 410–424.
[2] S. Li, X. Kang and J. Hu, Image fusion with guided filtering, IEEE Trans.
Image Process. 22 (2013) 2864–2875.
[3] A. P. James and B. Dasarathy, Medical image fusion: a survey of the state of
the art, Inf. Fusion 19 (2014), 14–19.
[4] R. Singh, R. Srivastava, O. Prakash and A. Khare, Multimodal medical
image fusion in dual tree complex wavelet transform domain using max-
imum and average fusion rules, J. Med. Imaging Health Inf. 2 (2012),
168–173.
[5] B. Yang and S. Li, Multifocus image fusion and restoration with sparse
representation, IEEE Trans. Instrum. Meas. 59 (2010), 884–892.
[6] M. Aharon, M. Elad and A. Bruckstein, K-SVD: an algorithm for designing
overcomplete dictionaries for sparse representation, IEEE Trans. Signal
Process. 54 (2006), 4311–4322.
[7] N. Yu, T. Qiu, F. Bi and A. Wang, Image features extraction and fusion
based on joint sparse representation, IEEE J. Sel. Top. Signal Process. 5
(2011), 1074–1082.
[8] H. Yin and S. Li, Multimodal image fusion with joint sparsity model, Opt.
Eng. 50 (2011), 067007.
[9] B. Yang and S. Li, Pixel-level image fusion with simultaneous orthogonal
matching pursuit, Inf. Fusion 13 (2012), 10–19.
[10] J. Wang, J. Peng, X. Feng, G. He, J. Wu and K. Yan, Image fusion with
nonsubsampled contourlet transform and sparse representation, J. Electron.
Imaging 22 (2013), 043019, 1–15.
[11] M. Nejati, S. Samavi and S. Shirani, Multi-focus image fusion using
dictionary-based sparse representation, Inf. Fusion 25 (2015), 72–84.
[12] H. Liu, Y. Liu and F. Sun, Robust exemplar extraction using structured
sparse coding, IEEE Trans. Neural Netw. Learn. Syst. 26 (2015), 1816–1821.
[13] M. Kim, D. K. Han and H. Ko, Joint patch clustering-based dictionary
learning for multimodal image fusion, Inf. Fusion 27 (2016), 198–214.
[14] N. Aishwarya and C. Bennila Thangammal, An image fusion framework
using morphology and sparse representation, Multimedia Tools Appl. 77
(2017), 9719–9736.
[15] Y. Zheng, E. A. Essock, B. C. Hansen and A. M. Haun, A new metric based
on extended spatial frequency and its application to DWT based fusion
algorithms, Inf. Fusion 8 (2008), 177–192.
 Adaptive dictionary-based fusion 137

[16] Q. Zhang and B. Guo, Multi-focus image fusion using the nonsubsampled
contourlet transform, Signal Process. 89 (2009), 1334–1346.
[17] Z. Hu, Medical image fusion using multi-level local extrema, Inf. Fusion 19
(2013), 38–48.
 Chapter 8
Artificial intelligence for sustainable e-Health
Shrikaant Kulkarni1

The study of exploring the improvement of efficiency in e-Health is done by means

of regulating an access to electronic health records (EHRs). In the absence of
appropriate apex bodies, EHR will continue to stay as lopsided and discrete network
of lagging systems without much ability to attain accuracy and consistency, and
thereby efficiencies. A multinational corporation (MNC) model is prescribed to cut
down health-care (HC) expenses and execute a coherent system wherein data, tech-
nology, and training are consistently upgraded to remove any interoperability-related
problems. The literature review reveals that EHR interoperability issues may be met
by generating architectures that drive fragmented systems to interoperate on the
guidelines of watch dog agencies. This chapter suggests a fundamental technology-
driven model predicting the need to get over interoperability problems, and followed
by suggesting an organizational model that would be the most suitable solution
catering to the needs of a coherent system where data, technology, and training are
consistently and regularly upgraded. Hence, an artificial intelligence (AI)-driven
model is prescribed to facilitate the improvement in the efficiency of e-Health to
standardize HER. This treatise deliberates on the research opportunities to provide
sustainable e-health solutions particularly during pandemic like COVID-19 and keep-
ing in view diabetes HC as a case study.

8.1 Introduction
This work is aimed at eliciting how an MNC organizational model can replace the
UK National Health Services (UK-NHS) government model, in situations wherein a
public-sector model is difficult to develop. Why and how the MNC model can offer
solutions of its own kind to a viable and well-articulated EHR system. It is prescribed
that the quality assurance of HC and efficiency with which it is accessed to EHRs can
be enhanced if right solutions are derived to the interoperability issue among insti-
tutions, organizations, and computer systems. HC is in fact an area where rather than
efficiency effectiveness is of paramount importance, the efficiency with which HERs
are accessed is key because of the sensitivity of medical decision-making to time [1].

1
Adjunct Professor, Faculty of Science & Technology, Vishwakarma University, Pune, India
140 The Internet of medical things

One of the reasons for inadequate efficiency is regulated access to EHRs. In

the absence of a supra organization, EHR systems will stay fragmented, leading to
nonuniform networks of lagging and degraded systems, or subsystems that are not
able to attain accuracy, consistency, and efficiencies in offering qualitative access
to EHRs. Currently a supra organization of such a kind is found named UK-NHS
(providing services to 6 crore people); few other nations too have alike organiza-
tions. However, these private or public umbrella organizations are not applicable
across the board, like USA, having a population of 31 crore people. The MNC-
based organizational model may be taken as a substitute to the nonprofit or for-
profit NGO or MNC reply. Moreover, this model is driven by AI in order to
facilitate EHRs interoperability on an e-Health platform [2].
This study defines EHRs and electronic patient records (EPRs), explaining the
nature of interoperability issue, underscores certain constraints in executing a
viable EHR system, and how to overcome these barriers. For addressing the
interoperability problems, the organizational model prescribed is a typical MNC
model associated with Internalization Theory, responsible for achieving effi-
ciencies, economies of scale, and cut down HC expenses, which would be intrin-
sically more because of fragmentation and problems of interoperability. It means
that the conclusions depending upon literature review are that most of the inter-
operability problems concerned with HER can be fulfilled by generating one or
more infrastructures and architectures that facilitate fragmented systems to interact
among themselves under supra organizations [3].

8.2 Reduction in margin of error in healthcare

HC is a service industry that expects abysmally low margin of error as against other
services [4]. In HC, an error would prove fatal, and cannot be reversed, similar to
an error from airline pilot, although redundant systems are inbuilt in an aircraft. In
fact duplicate systems in HC may not be encouraged due to cost effectiveness
problems, most important are accuracy, the framework of information architectures
within data flow highways to assure quality of data, and technology, constancy in
training and semantics for those feeding the data, information, and decisions. To
attain such goals, it is indispensable that an organizational structure that is an
umbrella for subsidiary units such as hospitals and HC units is required to coordi-
nate and acquire efficiencies in health records entry and providing in an accurate
and timely fashion. In the absence of such a supra organization, the subsidiary units
will stay discrete, or fragmented, leading to critical interoperability problems and
time lags in patient HC. The absence of suitable communication systems and
information sharing among HC peers and with patients is also considered as a
major challenge and is responsible for causing one in five medical errors due to
access to inadequate information which is a barrier in effective decision-making.
Another implication of fragmentation of computer systems is the duplication of
procedures, which enhances costs, apart from discomfort to patients [5]. Earlier
research shows that EPRs will check the medical errors to a large extent. However,
 Artificial intelligence for sustainable e-Health 141

studies have suggested that the efforts to employ EHR systems slow down proce-
dures as the interface is user unfriendly, although the interface was done with care
keeping in view professional users in perspective, and who evaluated it prior to use
on a mass scale. There is still no solution in sight that takes care of the optimum
utilization of latest information and communication technology (ICT) input/output
systems [6].
EHRs- and ICT-associated problems have come to the fore. American Senator
Hillary Clinton said “We have the highly sophisticated medical system in the
world, still patient safety is compromised because of medical errors, duplication,
and lack of efficiencies. Tapping the full potential of information technology will
check errors and better quality in our health system.” It is also found that electronic
medical records should be favored over chapter-based ones [7].

8.3 EPRs, EHRs, and clinical systems

According to UK-NHS, EPR is a periodic record of patient care supplied by an acute
hospital that provides HC treatment to the patient within a given time span, while EHR
is a patient’s life-long record consisted of numerous EPRs. According to International
Standards Organization (ISO), an EHR is a bank of patient-centric health-related
information that provides efficiency and integration of HC quality by offering plans,
goals, and assessments of patient care. EHR is the compilation of discrete data of a
patient [8]. A clinical record system should be put in place before EPR becomes
functional, which will result into EPRs over time. Right clinical systems will link
departments through a master patient index and the system will be integrated with an
electronic clinical result reporting system. A clinical system, comprising a secure
computer system, is a must for designing an EHR and EPR that enables hospital
computer systems to communicate among themselves and allows physicians to secure
patient data from various hospitals. The system in the entirety would be able to inte-
grate with selective clinical modules and document imaging systems to offer key
support. Generally, an EHR is enabled when state-of-the-art multimedia and tele-
medicine are hyphenated with a host of communication applications [9].

8.4 Barriers in an EHR system

Computer systems apart from fragmented or mutually exclusive cause ineffi-
ciencies, user unfriendliness, and obstructions to EHR systems in respect of [10]
● interfaces that demand betterment to better present HER;
● the way data is acquired;
● framing rules and regulations to acquire patient feedback and consent when
sharing their data on HER;
● technology issues to deal with EHR execution because of large data transfers,
their privacy supervision, and complexity of the system depending on existing
ICT infrastructures;
142 The Internet of medical things

● data quality, its availability and acceptability across nation by patients, medical
practitioners, and health workers is a must for EHR development; and
● one more major bottleneck in EHR application is that medical professionals are
not able to perform their jobs due to significant data entry needs to populate EHRs.

8.5 Getting over interoperability issues

Interoperability of EHR takes place when medical-record-enabled HC systems

(patients, medical practitioners, nurses, and concerned HC-department-centric) can
communicate among themselves. Interoperability is observed when one application
receives data from other application and then does useful tasks [11]. Moreover,
interoperability refers to IS that supports functions such as
● physical access to patient data;
● access in between suppliers of date in various care arrangements;
● access and conduct patient tests and medicines;
● access to computer-driven decision support system (DSS);
● access in a secured electrical-based communication among patients and
physicians;
● administrative processes like scheduling automated;
● access for a patient to disease management tools, patient records, or health-
related data;
● access to automated insurance claims; and
● access to database for safety of patients and efficiency in HC [12].
To a certain degree, interoperability issues have been removed by the ISO that
has created structure- and function-related EHR standards and processing systems.
The ISO has published 37 HC-related standards dealing with compatibility and
interoperability. These standards are divided into three subsections:
● EHR content, structure, and context wherein it is employed and shared;
● technical-architecture-related specifications for fresh EHR standards to
exchange EHR for designers called as open her; and
● standards to accomplish interoperability among HC applications and systems,
in addition to messaging and communicating criteria for designers to enable
reliable information transfer [13].
The interoperability issues have many bottlenecks, other than legal issues, ICT
and EHR protocols, IT architectures, training, constancy in advancing technologies,
entry accuracy, interpretation, etc. The organizational structure that tackles all
issues need not be a government agency, preferably working either for profit or no
profit such as UK-NHS and setup in nations like the United States. The objectives
and goals would differ from one geographical area to another, although the orga-
nizational model could be alike. In the United States, a nonprofit organization or
for-profit one would be acceptable similar to health insurance entities that are
private companies and have networks either at local or national level of HC servers
 Artificial intelligence for sustainable e-Health 143

and facilities; however, only the insurance companies, the medical practitioners or
the facility, and the patients are the repositories of their records and data. If an
individual leaves an insurance company but not the doctor, then the doctor can
make use of that EPR; however, on changing the doctor and if some other insurance
company is involved, then there are a host of protocols to be followed to get
records, which frequently are converted in terms of chapter records.
This happens mainly due to legal and privacy issues, and since there are non-
automated or computerized systems that communicate with one another for seamless
data transfer in between two doctors or facilities that are related to two different
insurance firms. The interoperability issue in such a situation can be addressed only
by proper coordination; however, another factor that often becomes a barrier is the
legal system that governs the entities. The legal issues can vary from reluctance to
share proprietary information, to leaking of information for either use or misuse of
the EPR by the legal system, the lawyers of the patient, or of another company. Thus,
fragmentation occurs not merely to the legal environment and the nonavailability of
linked data architectures but also because of an organizational structure, either public
or private, that can carry out all transactions with such a legal entity as like UK-NHS,
and alike entities in other nations, but a best option is a supra organization in the
NGO area or private sector would be encouraged preferably [14].

8.6 Barriers in EHR interoperability

The barriers in linking and executing EHR are as follows:
● Adaptability to novice systems and work procedure by clinicians.
● Costs toward HC savings, government requirements, and motivation.
● Networking vendors who are required to be forced to ensure interoperable
systems.
● Standards that are required to be set to ensure communication.
● Legislation too is a concern and hence the US government has set up laws to
regulate HC funding, which if breached would invite penal charges. Moreover,
laws are in place that prevents interoperability. Lack of communication technol-
ogy to promote interoperability, the United States is lagging from other countries
in respect of technology to enable interoperability, although it is leading in per
capita spending, with Germany ranked second, and France third [15].
● Lack of training, or proper implementation of the shared care concept which is
innovative in patient HC, involves an integration of HC by a host of hospitals
and doctors that demands well formulated and soundly designed interoperable
IT systems, which will reduce patient HC quality, however, if systems are
poorly designed [16].

8.7 UK-NHS model: characteristics and enhancements

This is one of the modern HC systems in the world which has dealt with a number
of issues. However, since it belongs to public sector, it has some advantages, as
144 The Internet of medical things

well as limitations. There is a lot to learn from such supra organization, works using
legislation with huge funds to spend. However, this model may be worth replicating
in the public sector, a private-sector organization could come up with a coherent
HC system, with or without the efficiency level attained by the UK-NHS. However,
the model cannot be implanted in all countries because of variations in financial
and legislative constraints prevailing in different countries.
The local and national NHS’s IT systems require further upkeep or replace-
ment of incumbent IT systems to
● integrate;
● implement innovative national systems;
● integrate NHS CRS with concerned HC-based products and services, like E-
prescription, which enhances patient care by minimizing errors in prescrip-
tions, redundancy of data, employee time, and cost; and
● provide right infrastructure (N3) to enable the NHS with smart network ser-
vices and higher broadband connections to better patient care protocols by
fetching patient care records from anywhere and anytime, thereby cutting
down HC costs by offering patient care remotely and not only saving time but
also by providing patient care expeditiously [17].

8.8 MNC/MNE characteristics

An alternative to UK-NHS organizational model has some characteristics that may
not be applicable to HC but do so with things like interoperability across the coun-
tries and jurisdiction. An MNC is any company that owns, controls, and manages
income creating assets across countries [18]. In this perspective, the parameters of
interest are control and management of assets, or HC facility, across jurisdictions.
Following are MNC’s representative characteristics as mentioned in literature:
● MNC’s are popular for their capability in transferring technology, inducing the
technology diffusion, and providing employee training and managing skill
development. It means that it is a provider of HC with an ability of incorpor-
ating and implementing innovative ICT’s and upskilling those concerned.
● They also have the ability to bridge the missing links in technology in between
the overseas investor and the parent economy [19].
● There is intensive coaching for providers for quality assurance, managerial
efficiency, and marketing. American MNEs emphasize upon formalization,
while European MNEs stress on socialization of structure and process.
● Internalization theory describes the presence and functioning of MNE/MNC
[20]. It adds up to comprehension of the boundaries of the MNE, interfacing it
with the peripheral environment, and its organizational structure internally.
Because of nonavailability of markets and inadequate contracts lead to
opportunistic behavior from others, the company substitutes external contracts
by direct ownership and internal hierarchies that promotes better transactional
efficiencies [21].
 Artificial intelligence for sustainable e-Health 145

● MNC—Internalization Theory has also been characterized in terms of company

structures and advances its technology. The theory states that as the business
transaction costs in other nations are more, an MNC can acquire both tangible
and intangible savings as well as efficiencies by undertaking most of the activ-
ities, confined to its own organizational structure. Creating a control and
accountability over both material and human assets protects against leakage in
processes and intellectual capital and drives the MNC to attain cost efficiencies
with the help of internal contractual accountability. If such activities are brought
about via open market (through various companies, providers, etc.), particularly
in different legal environments (such as states, nations, and cities) would open
up many costly issues for small-scale companies, hospitals, because of com-
pliance issues, and reliance on external agents having their own agenda.
● One more characteristic of an MNE is its transformation into an e-MNE,
wherein the cyberspace is a global network of computers connected through
high-speed data links and wireless systems and thereby enhancing national and
global governance. Can an e-MNE be defined as an organization that has
campuses in many countries and its management done through cyberspace?
[22]. Most of the cyberspace MNCs have attained economies of scale and have
capabilities or proficiency in cutting down costs.
● Currently the e-MNE is capable of controlling and managing income creating
assets in many countries through a network across the world with an electronic
base situated in every building or place [23].
According to the internalization theory, two parallels can be elicited: one refers
to the circumstances and environment of the organization (as shown by UK-NHS
model), and the other is external market (as system variations exhibited in inter-
operability issues). The UK-NHS is an umbrella for over 6 crore people as a supra
organization which controls and accountable to the UK government legal autho-
rities. The interoperability issues in United Kingdom are not within the ambit of
legal jurisdictions, but in technology, data architectures, and human issues.
Jurisdictional matters come to the fore when one shifts from the United Kingdom’s
legal environment, to USA, or other countries not at par.

8.9 UK-NHS model or the MNC organizational model

An EHR system tries to accomplish what the industries have already achieved. HC
is complex and, thus, asks for a customizable model to fulfill its own and patient’s
needs. Moreover, lack or inadequate information in EHR prevents clinicians in
making quality decisions. Hence, more efforts are to be taken for coordinating input
and output. Therefore, an organizational model and structure fits into an environ-
ment where interoperability issues can be addressed when faced having two or
more complex systems [24]. The UK-NHS is such a kind of organizational model
that tries to get over interoperability problems via its writ of legislation and the law,
which can enacted as it is a government organization. However, in spite of a
healthy legal and political environment, there are host of interoperability problems
146 The Internet of medical things

that exist because of technology, training, and behavioral resistance. An alternative

solution to the UK-NHS model depends primarily on an organizational model
designed by MNC over many years and is relevant because they work across
numerous boundaries and legal systems. Earlier literature on MNCs show that
although these are private organizations working across several countries, they
have to handle various types of interoperability problems and consequently have to
get over the hurdles, which is attributed to their ability to address problems through
access and deployment of huge resources. In addition, such MNC model can be put
into effect in the e-Health sector enabled by AI.

8.10 e-Health and AI

Computing machines have transformed the HC sector in many dimensions, e.g.,
Internet of Things (IoT) [25] along with machine learning (ML) and AI as key
technologies. The function of AI is expanding in its horizons with its application to
the HC sector and has been emerging AI on the e-Health platform. AI is very vital
as datasets and resources are readily available. AI is already working for the cause
of HC sector: e.g., dermatology, oncology, radiology as representative examples.
AI and ML are considered as support tools for knowledge-driven medical decision-
making in patient care. AI is presently used where e-Health platforms transfer
patient data on integration, e.g., EHRs obtained in different environments, e.g., in
the patients’ homes and clinical warn room too [26]. This is a novel and sound
management information system that develops an AI-based DSS employable on
e-Health platforms.

8.11 Sustainable healthcare

ANH-USA defined sustainable health as “A complex system with interacting
strategies for restoring, managing and optimizing human health having an ecolo-
gical foundation, environmentally, economically and socially viable for indefinite
time, and works in a harmonious manner with both human and non-human envir-
onment, and which does not lead to unbalanced effects substantially on any factor
of the HC system” [27]. Present COVID-19 pandemic, which has derailed the
world crippling the best HC systems under its influence, directs strongly to all HC
systems to be connected with sustainability principles and asks for a major shift in
HC approach by nations for the betterment of society. Now it is time for the countries
to put into practice sustainable HC for their people. Conventional and alternative
medicines like Homeopathy, Ayurveda, Yunani, Chinese and Naturopathy about
which questions were raised for their scientific basis by the practioners of modern
medicines like Allopathy.
However, the alternative forms of medication have proved their potential
particularly during testing times and are in practice since long. It requires strong
desire to obtain, use, and analyze the data regarding conventional medication forms
and its usefulness employing AI/ML tools.
 Artificial intelligence for sustainable e-Health 147

8.12 Sustainable healthcare in the aftermath of COVID-19

Sustainability conceptually has benefited a lot various domains of business like
energy, agriculture, forestry, construction, and tourism. It is gaining momentum in
the advanced HC system which is prevalent in pharmaceutical drugs and products
[28]. However, at times it has been proved time and again, that the contemporary
medication was not effective against many infectious and chronic ailments.

8.13 Sustainability in staff and clinical practice during

pandemic

To attain sustainability in terms of faculty and clinical practice particularly during

pandemic situations like COVID-19, following recommendations are made:
● Tele-health technology helps clinicians to oversee patients at home and provide
recommendations for treatment, a robust infrastructure for telemedicine is asked for.
● COVID-19 although has effectively lower death rate than SARS or MERS, still
the pressure on HC systems across the globe is quite alarming [29]. The HC
systems should have sound plan to protect HC employees from infection and
fatigue during prolonged battle against COVID-19.
● Different HC offices are limited because of the lockdown timings, commu-
nication with policy makers as they formulate programs to provide necessary
government relief and timely support.

8.14 Broadening health-care facilities at home during

the COVID-19 pandemic

The HC environment can be expanded at home by using the following means:

● Hospital-at-home program can cut down HC expenses and mortality rate too.
● Hospital care-at-home can minimize the transmission risk of COVID-19, in
particular for patients who are vulnerable.
● Hospital-at-home to be effective, interdisciplinary social and behavioral health
services are demanded.
● Hospital-at-home and primary health services must include social and beha-
vioral health necessities too.

8.15 Sustainable development groups

Group of experts can be formed for the development of sustainability in HC as well

as in dealing efficiently and effectively pandemic situations in the following ways:
● The world wants to assure peace and prosperity for the mankind through col-
laborations (Governments–Private–NGOs–CSOs–Individuals) in all spheres.
COVID-19 pandemic should be used for the cause of sustainable development
in tandem with the SDGs.
148 The Internet of medical things

● Medical council should carry on with furtherance of primary and hospital-at-

home services in far flung areas too. The patients and HC teams should expand
their services both before and post pandemia.
● The technical support is meant for long-term initiatives so as to facilitate heath
service managers to maintain consistency and resource allocation during pan-
demic situations. This will assure that people carry on with care seeking
adhering to public health recommendations [30].

8.16 Futuristic research directions

ML can create new opportunities for sustainable HC and the investigators can focus
on automation in the following areas:
● analysis and prediction of COVID-19 cases;
● discovery of COVID-19 patients;
● current consolidated portal to support pandemics in the future;
● setting up of an innovative Pilot COVID-19 Data Warehouse for reference in
future;
● upgraded tools for collecting, preparing, and storing data thoroughly; and
● platform to support for building up a community of medical professionals for
pandemic situations.

8.17 Role of AI in diabetes care—a case study

AI is considered as a way wherein computers perform tasks that would otherwise
demand human intelligence. Diabetes mellitus is a chronic, pervasive state, data
rich and with a host of potential results. Therefore, diabetes is a strong case for
introducing AI. Clinicians at times pose the following questions:
● How will a diabetic patient respond to a given medication?
● Who will be vulnerable to acquire diabetic retinopathy?
● How can we create a glucose control algorithm for developing a safe closed
loop system?
● Which patients of mine are more vulnerable to influenza this fall?
● Of those with prediabetes, who are most likely to acquire diabetes and thereby
get benefitted from preventive measures?
Most of the people are part of a vibrant ecosystem, where one’s social,
biological, and other variables are available to interested ones. Although priv-
acy regulations regarding the access to such information by third parties will be
put into place beyond doubt, once an individual allows to share his or her data,
AI will result into better recommendations for a given individual. With
advancements in ML and deep learning (DL) accompanied by enhanced natural
language processing, AI should provide timely, quality, and validated advice
[31].
 Artificial intelligence for sustainable e-Health 149

The journey to HC future will cover advancements primarily in three areas:

1. HC technology,
2. HC delivery, and
3. computer science and information technology.
Advancements in HC technology cover breakthroughs in human genome pro-
ject, pharmaceuticals, nutraceuticals, and medical devices. Novel methods in the HC
have been devised for better disease control, evidence-enabled HC, and mind–body
medicine. Novel computer science and information technology is facilitating to deal
with and understand large amount of health information. Such developments are
taking place in terms of speed, storage capacity, mobile personal computing and
communication devices, cloud computing, AI, networking, and biometrics. The
Internet, which is cutting across many networks in a broad spectrum of frontiers, has
a substantial effect on all aspects of our lives. The generations to come of networks
will make the use of a plethora of resources with substantial sensing and intelligent
capabilities. These networks will cross physically connected computers to contain
multimodal information right from biological, cognitive-to-semantic, and social
networks. This major transition will have symbiotic networks of smart medical
devices (implantable, injectable, on-body), and smart phones and communication
devices. Such devices and the network will be consistently sensing, monitoring, and
interpreting the environment, which is at times called as the IoT [32]. The IoT and
social networks symbiosis is called the Internet of everything which will have major
impacts in the manner HC is offered. The P4 medicine concept coined by Leroy
Hood, Institute of Systems Biology, has been extended to the P7 concept by Sriram
and Jain in 2017. P7 concept for the HC future covers the following elements:
1. Personalized: Personalized medicine consists of individualized treatment.
2. Predictive: Depending upon the information in the EHRs and genomic data, an
individual’s vulnerability to specific diseases can be determined.
3. Precise: On collecting data and information, the analytical tools can be employed
to determine the cause of a disease and to suggest right therapeutic initiatives.
4. Preventive: Rather than treating a disease, ML and decision analytical tools
can be made use of in employing strategies to prevent the start of disease.
5. Pervasive: The HC should be made available at all time, everywhere, and at all
places.
6. Participatory: The patient should be actively involved in the diagnosis and the
treatment.
7. Protective: Proper safety measures should be adopted to make sure that the
patient data remains confidential.

8.18 Artificial intelligence and its area of applications

AI is related with ML, artificial neural network (ANN), DL, and knowledge-based
systems as shown in Figure 8.1. AI is used in addressing different types of issues
that surface in HC.
150 The Internet of medical things

Artificial intelligence (AI)

Knowledge
Machine learning (ML) -based
Systems
Artificial neural
networks (ANN)

Deep learning (DL)

Figure 8.1 Relation between AI, machine learning (ML), artificial neural network
(ANN), deep learning (DL), and knowledge-based systems (KBS)

8.18.1 Diagnosis
The diseases are made out depending upon potentially noisy data. The clinicians
should be able to correlate the symptoms to the right disease(s). Such a task
involves reasoning with inadequate and inexact data, defective sensors, etc.
Diagnosis is normally followed by a treatment plan, which involves determination
of a sequence of actions to treat the disease [33].

8.18.2 Interpretation
The collected data are analyzed to interpret. The data at times are unreliable,
erroneous, or extraneous. Therefore, the system should have capability to remove
outliers in the information.

8.18.3 Monitoring
Signals are constantly interpreted and alarms are put into effect as and when
needed.

8.18.4 Control
Signals or data are interpreted, and the system is regulated, depending upon
abnormalities in intended response [34].

8.18.5 Treatment planning

A program of actions is prepared to attain some goals. The actions initiated should be of
such type that surplus resources are not consumed and with no violation of constraints.

8.18.6 Drug design

Drugs that fulfill specific requirements can be designed. Such type of problem
covers meeting constraints from a host of sources. Design problems are normally
 Artificial intelligence for sustainable e-Health 151

addressed by breaking them into number of smaller ones. The designer must have
ability to handle the relation between these smaller tasks appropriately [35].

8.19 Conclusion

The AI systems first of its kind were knowledge-enabled DSSs, using isolated static
datasets. Such systems on connecting to electronic health data, AI became fruitful
for teaching, though not prepared for clinical care. DL via multilayered connected
ANNs has potential clinical applications. Google teams and New York University
observed that DL models can improve accuracy in lung cancer diagnosis. Although
our expectations of AI in clinical medicine and in diabetes are heightened, there
have been incremental developments showcasing the applicability of AI.
AI can be useful in developing synthetic pancreas (closed-loop systems) and
in comprehending the relation between social parameters and physiological bio-
markers of health. Diabetes too produces a significant amount of retinal, renal,
the taxonomy generated for diabetes with the help of root and rule-based method.
End-stage renal disease, vascular, and other data are monitored over the time.
Such pathophysiological data with other data will give new insights into diabetes
management. An expert opinions and care can be accessed to patients from any
part of the globe. A diabetes patient will have an access eye care from anywhere
through the application of smart phone images. Glucose control algorithms can be
individualized for the patients depending upon glucose consumption, insulin
pharmacology, nutria-genomics, and exercise patterns. It is clear that we are on
the verge of revolutionizing HC, with the ability to extract, analyze, the data for a
timely and quality clinical advice. AI will play a key role in bringing about a
paradigm shift in HC. It demands rapid progress in knowledge representation and
reasoning.
The UK-NHS model has proved that it can manage HC of 6 crore people, by
overcoming the interoperability issues and the countries that follow such public-
sector-governed HC system can adopt the model, if they possess strong political
and economic will power to follow. Countries, constitutions, legal and political
systems, financial condition, etc. of which if are not suitable in employing a UK-
NHS, or similar kind of model, can opt for an alternative in the form of MNC
model irrespective of whether it is developed as a nonprofit NGO or a for-profit
organization. Government sector organization gets the support from government
and its legislators to pass laws that enables the working of an HC, EHR, and EPR
system where interoperability issues are required to be made out and are to be
overcome by the legislature. In private-sector corporation, the interoperability
issues can be overcome by the generation of an internalized market under the aegis
of an NGO or a corporation.
An MNC organizational model was designed in order to get over various inter-
operability issues among countries. Therefore, an MNC model, having own inter-
nalized market to control, is a fit case in overcoming EHR interoperability issues,
integrating the internally related IS architectures, upgrading them, and training the
152 The Internet of medical things

workers with certain consistency. Despite this nonuniformity in HC software appli-

cations throughout EHR systems may still be a problem concern [36].
Further, MNC model has dealt with software, privacy, jurisdictional, and a host
of other issues in the finance while dealing with largely confidential information in
financial matters, and offering people access worldwide to accounts of them.
Therefore, while problems are not much complex, the HC sector is complex to a
greater extent as it involves not mere card swipe and data recording, but substantial
volume of subjective interpretations and inferences are drawn by HC service pro-
viders of plethora of skills and then transferred it to other HC service providers.
Ultimately, it was brought to the notice that the UK-NHS model and the MNC
model operate by legislating laws, and the MNC model by reaching contracts with
people, who are held accountable through the legal system. Additionally, AI is
incorporated to lay emphasis on its application within the e-Health platform, so as
to facilitate the proposed MNC model across the globe. Making out the knowledge
hidden in the data is a quite challenging job in ML. This chapter focuses on the
measures to be taken for the sustenance of e-Health and can be employed for the
better HC of diabetic patients as well as during and post pandemic crisis.

References
[1] H. Almubarak, R. Stanley, W. Stoecker, and R. Moss, Fuzzy color clustering
for melanoma diagnosis in dermoscopy images. Information 8(89) (2017).
[2] A. Angulo, Gene selection for microarray cancer data classification by a
novel rule-based algorithm. Information 9, 6 (2018).
[3] Avon Health Authority, Electronic Patient Records Electronic Health Records,
Avon Health Authority, Schofield, J, Bristol, Bristol 15, 389–406 (2000).
[4] A.R. Bakker, The need to know the history of the use of digital patient data,
in particular the EHR. Int. J. Med. Inf. 76, 438–441 (2007).
[5] C.A. Bartlett and S. Ghoshal, Managing Across Borders: The Transnational
Solution, Harvard Business School Press, Boston, MA (1989).
[6] B. Baumann, Polarization sensitive optical coherence tomography: a review
of technology and applications. Appl. Sci. 7, 474 (2017).
[7] A. Begoyan, An overview of interoperability standards for electronic health
records. In Society for Design and Process Science. 10th World Conference
on Integrated Design and Process Technology; IDPT-2007. Antalya,
Turkey, June 3–8 2007.
[8] K. Chung, R. Boutaba, and S. Hariri, Knowledge based decision support
system. Inf. Technol. Manage. 17, 1–3 (2016).
[9] Commission on Systemic Interoperability, Ending the Document Game, U.S
Government Official Edition Notice, Washington, DC (2005).
[10] R.C. Deo, Machine learning in medicine. Circulation 132, 1920–1930
(2015).
[11] J. Dunning, Multinational Enterprises and the Global Economy, Addison-
Wesley, Wokingham (pp. 3–4) (1992).
 Artificial intelligence for sustainable e-Health 153

[12] S. Garde, P. Knaup, E.J.S. Hovenga, and S. Heard, Towards semantic

interoperability for electronic health records: domain knowledge governance
for open EHR archetypes. Methods Inf. Med. 11(1), 74–82 (2006).
[13] I. Guyon, J. Weston, S. Barnhill, and V. Vapnik, Gene selection for cancer
classification using support vector. Mach. Learn. 46, 389–422 (2002).
[14] A. Harrison, The role of multinationals in economic development: the ben-
efits of FDI. Columbia J. World Bus. 29(4), 6–11 (1994).
[15] D. Impedovo and G. Pirlo, Dynamic handwriting analysis for the assessment
of neurodegenerative diseases. IEEE Rev. Biomed. Eng. 12, 209–220 (2018).
[16] S.M. Islam, D. Kwak, M.H. Kabir, M. Hossain, and K. Kwak, The Internet
of Things for health care: a comprehensive survey. IEEE Access 3, 678–708
(2015).
[17] A. Jalal-Karim and W. Balachandran, The influence of adopting detailed
healthcare record on improving the quality of healthcare diagnosis and
decision making processes. In Multitopic Conference, 2008 IMIC. IEEE
International, 23–24 (2008).
[18] A. Jalal-Karim and W. Balachandran, Interoperability standards: the most
requested element for the electronic healthcare records significance. In 2nd
International Conference–E-Medical Systems, 29–31 Oct 2008, EMedisys.
IEEE, Tunisia (2008).
[19] A. Kokko, Technology, market characteristics, and spillovers. J. Dev. Econ.
43(2), 279–93 (1994).
[20] Y. Li and L.S. Shen, lesion analysis towards melanoma detection using deep
learning network. Sensors 18, 556 (2018).
[21] A. Massaro, V. Maritati, N. Savino, et al., A study of a health resources
management platform integrating neural networks and DSS telemedicine for
homecare assistance. Information 9, 176 (2018).
[22] NHS National Program for Information Technology, Making It Happen
Information About the National Programme for IT. NHS Information
Authority, UK (nd), Michigan.
[23] W.P. Nunez, Foreign Direct Investment and Industrial Development in
Mexico, OECD, Paris (1990).
[24] A. Razzaque and A. Jalal-Karim, The influence of knowledge management
on EHR to improve the quality of health care services. In European,
Mediterranean and Middle Eastern Conference on Information Systems
(EMCIS 2010). Abu-Dhabi, UAE (2010).
[25] A. Razzaque and A. Jalal-Karim, Conceptual healthcare knowledge man-
agement model for adaptability and interoperability of EHR. In European,
Mediterranean and Middle Eastern Conference on Information Systems
(EMCIS 2010). Abu-Dhabi, UAE (2010).
[26] A. Razzaque, T. Eldabi, and A. Jalal-Karim, An integrated framework to classify
healthcare virtual communities. In European, Mediterranean & Middle Eastern
Conference on Information Systems 2012. Munich, Germany (2012).
[27] A. Razzaque, M. Mohamed, and M. Birasnav, A new model for improving
healthcare quality using web 3.0 decision making. In Making it Real:
154 The Internet of medical things

Sustaining Knowledge Management Adapting for Success in the Knowledge

Based Economy ed. by A. Green, L. Vandergriff, A. Green, and L.
Vandergriff, Academic Conferences and Publishing International Limited,
Reading (pp. 375–368) (2013).
[28] A.M. Rugman, Inside the Multinationals: The Economics of Internal
Markets, Columbia University Press, New York, NY (1981) (Reissued by
Palgrave Macmillan 2006).
[29] O. Saigh, M. Triala, and R.N. Link, Brief report: failure of an electronic
medical record tool to improve pain assessment documentation. J. Gen. Int.
Med. 11(2), 185–188 (2007).
[30] I. Sluimer and B. Ginneken, Computer analysis of computed tomography
scans. IEEE Trans. Med. Imaging 25, 385–405 (2006).
[31] D. Stacey, F. Légaré, K. Lewis, et al., Decision aids for people facing health
treatment or screening decision. Cochrane Database Syst. Rev. 4, CD001431
(2017).
[32] O.E. Williamson, Markets and Hierarchies, Analysis and Antitrust
Implications: A Study in the Economics of Internal Organizations, Free
Press, New York, NY (1975).
[33] G. Zekos, Foreign direct investment in a digital economy. Eur. Bus. Rev. 17
(1), 52–68 (2005).
[34] M. Rigla, I. Martı́nez-Sarriegui, G. Garcı́a-Sáez, et al. Gestational diabetes
management using smart mobile telemedicine. J. Diabetes Sci. Technol. 12
(2), 260–264 (2018).
[35] D. Grady, A.I. Took a Test to Detect Lung Cancer. It Got an A. New York
Times, New York (2019). Available at: https://www.nytimes.com/2019/05/
20/health/cancer-artificial-intelligence-ct-scans.html. Accessed: May 26,
2019.
[36] G. Marcus and E. Davis. Rebooting AI: Building Artificial Intelligence We
Can Trust. Pantheon, New York, NY (2019).
 Chapter 9
An innovative IoT-based breast cancer
monitoring system with the aid of machine
learning approach
Bichitrananda Patra1, Santosini Bhutia1, Trilok Pandey1
and Lambodar Jena2

The increasingly revolutionary Internet of Things (IoT) grows rapidly in modern

life with the intent of increasing the quality of life by integrating a variety of
intelligent tools, technologies, and applications. The IoT Medical Devices will
revolutionize the medical industry by creating an environment where patient data is
transmitted to a cloud-based storage, processing and analysis network. This chapter
provides novel tools for the continuous follow-up of breast cancer (BC) patient
conditions—and iTBra that has incorporated sensors that monitor cell temperatures
and transfer data on to a patient database in real time. An abnormal reading causes
an alert to the patient and his doctor via smartphone. In reality, it is a big obstacle
for medical professionals and researchers to detect BC early in their lives. In order
to address the initial phase finding issue of BC has projected a machines-built
diagnostic instrument that would identify malignancy and benign into IoT. For the
implementation of our proposed method, a machine learning sequential minimal
optimization (SMO) classification is used to distinguish malignant and benign of
healthy citizens. We use support vector machine- recursive feature elimination
(SVM-RFE) rank approach to improve the classification performance of BC sys-
tems by selecting all relevant features from the datasets. For preparing datasets, the
process for training and evaluating the ranking integer aimed the finest prophetic
model. In addition, concert assessment classification measures such as classified
value of TP rate, FP rate, precision, recall, Matthews correlation coefficient
(MCC), and ROC area were used to track classified results. The dataset “Wisconsin
Diagnostic Breast Cancer (WDBC)” was employed in this study to check the pro-
posed procedure. Experimental results show that the repeated algorithm for the
choice of features selects the best subset and classification instead SMO achieves
the highest result. In addition, we are proposing this system to diagnose BC

1
Department of Computer Science and Engineering, Siksha ‘O’ Anusandhan (Deemed to be University),
Bhubaneswar, India
2
K L Deemed to be University, Vaddeswaram, Andhra Pradesh, India
156 The Internet of medical things

effectively and in efficiently early stages. With this program, the reconstruction and
treatment of BC would be more successful. Controller-assisted devices transmit all
sensor values as packets to the mobile app via Bluetooth. We predicted and alerted
the shift in patient status with the aid of the machine learning algorithm.

9.1 Introduction

BC has been the world’s utmost severe and wide-ranging syndrome, with 2 million
new cases occurring during 2018, according to the American Research Institute [1].
By improving clinical outcomes, and increasing cost-effectively patient care, the
health-care sector contributes to two key objectives. In order to increase clinical
results, real-time measurement and treatment of diseases are considered key fea-
tures. Patient care needs to be improved to improve patients’ quality of life and user
experience. IoT offers a cost-efficient way of achieving the two objectives for the
medical devices health industry. Connected health programs use scarce resources to
enhance health quality, which results in better clinical results. Measurable asso-
ciated advantages. Medical devices include reduced death rates, less visits to hos-
pitals, and less medical devices.
Hospital and ambulance admissions include reduced bedding and hospital stay.
Remote patient control results inefficient and prompt diagnosis and leads to
improved health-care management [2]. In addition, patients (and their relatives)
may become more visible and actively involved, by making them more visible in
their current health conditions.
BC cells in BC tissue are growing abnormally and the rate of cell diminution,
resulting in BC, is slowly increased. Typically, BC is a malignant cancer in the
breast. Most dividing cells are tissues known as cancer (malignant) and non-
cancerous (benign). Tumors are a piece of or build of tissue. In advanced medical
research development nations, 5-year primary BC survival rate is 80%–90%,
though secondary BC survival rate falls to 24% [3]. Specific invoice-based methods
were used to diagnose BC. Breast tissues are collected for examination in the
biopsy technique [4] and the result is highly accurate. Nevertheless, the patient has
discomfort in taking a biopsy from the breast. Mammograms [5] are also used to
treat BC and are used for BC diagnosis. Yet diagnosis of benign cancer is not easily
rendered by the mammogram technique. Magnetic reasoning imaging [6], a very
complex method that offers excellent outcomes with three-dimensional (3D) ima-
ges and shows interactive features, is also an invoice-based technique for diag-
nosing the breast. Such methods for invoice-based diagnostics are highly difficult
to implement and the tests do not diagnose BC efficiently and reliably.
Wearable technology offers continuous surveillance of BC women, unlike
traditional invasive mechanisms. Basically, two solutions exist: traditional methods
invasive in nature and one is the noninvasive method more frequently used. The use
of IoT on devices like iTBra simplified monitoring for BC even further. Such IT
devices monitor changes in a woman’s metabolic profile overtime and distinguish
between healthy and potentially cancerous profiles. The use of metabolic detection
 An innovative IoT-based breast cancer monitoring system 157

represents a global breakthrough in cancer screening over image-based technolo-

gies such as mammography or ultrasound [7]. In addressing these global health
problems, a cost-efficient and time-consuming end-to-end approach needs to be
given. A new, noninvasive approach such as the machine learning technique is
more efficient and accurate in addressing these problems in invasive approaches for
diagnosing BC. Machine learning methods were used in the literature to distinguish
breast tissues these are either malignant or benign. This research briefly addresses
the related literature on machinery for BC diagnosis.
The solution includes a method that measures signs like radius, textures, the
perimeter, the area, smoothness, compactness, concavity, concave points, sym-
metry, and the fractal dimension in the proposed study. Such signs are important to
any patient and the collected data will be used to avoid the patient’s sudden
increase/decrease. The main goal of this chapter is toward provide an IoT-built
method of prediction grounded on machine learning for the successful analysis of
patients with BC besides healthy individuals. The predictive master education
SMO classical was applied to the classification of BC in malignant and benign
persons. For selecting features improving SOM classification efficiency, the SVM-
RFE feature selection (FS) algorithm has been adopted. In this chapter, we adopted
SVM-RFE for correct selection because SVM-RFE FS-based system classification
performance is good in comparison to other BC and safe people classification
methods.
We begin with a brief overview of IoT on the iTBra framework and machine
learning support for BC in Section 9.2, and Section 9.3 describes the iTBra concept,
dataset preprocessing, SVM, SVM-RFE, SMO, and reforms of SMO algorithms are
introduced with the BC forecast proposed predictive framework. Section 9.4
explains the experimental frame and performance. Finally, Section 9.5 introduces
the conclusion and potential research.

9.2 Related works

As a new device, iTBra, demonstrates, the impact of IoT devices continues to grow.
Associated bra is built for the early detection of BC, which will be put on sale
globally in the first half of 2018 by Rob Royea [8]. Whereas mammograms are the
primary form of recognizing BC, cancer cells in dense breast tissue are more dif-
ficult to detect as they have more tissue and less fat. According to Susan G. Komen,
40%–50% of females in the United States aged 40–74 have thick breasts. The
iTBra from Cyrcadia Health comprises 16 sensors for breast tissue adjustments.
The sensors are attached to the skin of the patient with a screw. The patient uses the
iTBra for 2 h, and the information gathered is sent to the doctor for review. It is an
alternative to the pain of mammograms, particularly for people with dense breast
tissues like Royea’s wife, Kelli Royea, who starred in the film. Royea has
explained that the bra operates “through the prediction method of the algae” (arti-
ficial intelligence), all research is done at a cloud-based analytical venue. Screening
tests are immediately sent to the customer, showing whether they are successful or
158 The Internet of medical things

you may want to see the doctor. The purpose of Royea was to make this established
technology a portable product that could be sold as a wearable tool, so that millions
of women could easily wear it and build an overall analytical database for early
detection. This is modified by the iTBra. “Rob and his company have been able to
attach sensors on the breast in search of cell-level changes in cancer and don’t have
to be seen by a doctor with his eye,” explained Kramer.
This structure has their own advantages, but do not have an early symptom
identification solution in terms of real-time predictions. The aim of this chapter is,
therefore, to overcome this restriction in order to create a complete system that
measures all BC characteristics necessary. In addition, the patient may be alerted
before the disease worsens, thus decreasing the women beings mortality rate, by
using the prediction algorithm on the real-time data. This inactive alternative is a
good example to constantly track and avoid patient complications when combined
with the real-time information allows prediction by the SMO classification with
SVM-RFE FS and allows recommending suitable remedies.
Azar and El-Said [5] have suggested a BC screening technique. They
employed three classification methods: radial basis function kernel (RBF), prob-
abilistic neural network (PNN), and multilayer perceptron, respectively. The results
of the classifiers were evaluated by applying performance assessment methods,
including precision, specificity, and sensitivity. Precisely 97.80% and 97.66% of
training and assessment classifications, respectively, were obtained and the accu-
racy was achieved. In addition to genetics, AliăźKović and Subasi proposed the two
Wisconsin brain cancer datasets [9].
Using the fuzzy-genetic algorithm system that hits 97.36% accuracy, Akay [10]
suggested using the F-score system for BC screening and endorsing the vector
machine (SVM) and had strong output results. The K-means Zheng et al. [11]
algorithm used for the aggregation and extraction properties and for the classification
of brain and malignant breast tumors in combination with SVM. In classification and
low processing time, the technology proposed achieved great precision [12]. Used in
conjunction with various classifiers for Ramadevi Hybridized Primary Component
Analysis (PCA) used in many datasets on BC and realized fair precision. Now [13]
the writer suggested a Pareto-based mimetic strategy, the Naive Bayes Classifiers
(NBC) Detection Network. The experimental results have shown that the technology
proposed has achieved reasonable grading precision and very low machine time.
For feature extraction, the K-means and SVM were used for classification.
Intelligent BC detection technique was developed by Onan [14]. For example, the
floppy-through was used to pick and select functions based on consistency. In BC
screening, he recycled the nearest neighbor algorithm. The technology for the opti-
mization of swarm particles and nonparametric estimates for the kernel density of the
BC density was developed by Sheikhpour et al. [15]. Rasti et al. [16] have developed
a diagnostic technique of BC using a combination of convolutionary neural networks
and have achieved 96.39% precision. Ani et al.’s IoT-built treatment and control [17]
93% accuracy has been reached by the ensemble classifier and system prediction
process. Yang et al. [18] proposed a wearable ECG cloud-based intelligent IoT health
system and proposed a very good system results for disease diagnosis.
 An innovative IoT-based breast cancer monitoring system 159

This chapter’s main goal is to implement an IoT-based approach for fore-

casting BC patients and healthy individuals by computer. In the BC classification
of malignant and benign individuals, the predictive SMO model in master
learning has been used. For selecting features improving SMO classification
efficiency, the SVM-RFE FS algorithm has been adopted. In this chapter, we
adopted SVM-RFE for correct selection because SVM-RFE FS-based system
classification performance is good in comparison to other BC and safe people
classification methods.

9.3 Proposed solution

9.3.1 Concept of iTBra
The gadget is covered with a woman’s dress. Thermodynamic assessment will allow
sensors that sit on the body, explain Royea [19], and experience metabolic change
over time. Finally, he claims that the earliest symptoms of invasive cancer may be
identified. The results are sent back to patients, the doctor, and the insurance provi-
der. It is simple for patients to read results which essentially tell them “you’re fine”
or “call your doctor.” The data is transmitted to a central laboratory in which artificial
intelligence or predictive analysis is used in the cloud for determining the results.
Therefore, the iTBra appears to be a very early warning that may be essential
to the cancer rather than a mammogram that cannot locate a tumor for years. The
iTBra is a clinical trial in the United States with a clearance of FDA Class 2, but
Royea claims that it operates on a clearance of Class 1, and it can be marketed over-
the-counter. A company product could meanwhile be available in Asia by the end
of the year, allowing iTBra to combine with a smartphone for performance.

9.3.2 Dataset
The data collection “WDBC” obtainable on machine learning source [20] has been
developed by Dr William Wolberg from the University of Wisconsin. This was
used as a dataset for the conduct of the proposed research to develop master
training programs for diagnosing BC. The dataset includes 569 items, 32 attributes,
and 30 characteristics are real-value characteristics. The diagnosis on the target
performance label has two classes for the malignant or benign individual. Total 357
benign and 212 malignant individuals are classified in class. Therefore, the dataset
is a matrix of 569
32.

9.3.2.1 Preprocessing dataset

Information processing is required before applying the classification algorithms.
The data processed [21] decreased classifier processing time and the efficiency of
the classifier improved. The dataset is commonly used for methods like defective
value recognition, regular scalar, and min–max scalar. Standard scalar guarantees
the mean 0 of each function and the variance; thus, all features have one coeffi-
cient. For all functions, min–max scalar moves the data to 0–1. The vacant free
function in the row is excluded from the dataset.
160 The Internet of medical things

9.3.3 Feature ranking with support vector machines

9.3.3.1 Support vector machines
We recycled an advanced classification technology to check the concept of using
the sizes of a classifier to construct a function ranking: vector support engines
(SVMs) [22]. SVMs have been researched and benchmarked intensively in recent
years for a variety of techniques. Actually it is one of the most popular strategies
for machine classification of their rivals.
While SVMs deal with arbitrary complexity’s nonlinear decision limits, linear
SVMs are common linear classifiers discriminating. A linear SVM is an extreme
edge classifier for training dataset that is linearly separable. The decision-making
boundary (a straight line for a level separation) is positioned so that the maximum
distance on each side can be left. One advantage of SVMs is that the weights wi of
the decision function D(x) are only a small subset function of the “support vectors”
training examples. These are the cases nearest to and on the edge of decision-
making. The presence of such supporting vectors contributes to the computational
properties of SVM and its efficient success in classification. Although SVMs base
their decision function on borderline support vectors, they also base their decision
function in the average case on other methods, including the approach used in the
study of Guyon et al. [22].
Efforts: Preparation {x1, x2, . . . xk, . . . xl} and class labels {y1, y2, . . . yk, . . . yl}.
Reduce done ak :

X X
1
J¼ yh yk ah ah ðxh  xk þ ldhk Þ  ak
2 hk k
X
Subject to: 0  ak  C and ak yk ¼ 0
Productivities: Constraints ak .k
The resultant assessment function of an input vector x is
DðxÞ ¼ w  x þ b
X
with w ¼ ak yk xk and b ¼ ðyk  w  xk Þ
k

9.3.3.2 SVM recursive feature elimination

A function selection process can be used to remove terms that are statistically
unrelated to class labels from the training dataset, thereby improving both effi-
ciency and precise. The supervised method of dimensional reduction was provided
by Pal and Maiti [7]. A mixed 0–1 integer program was used for the feature
selection problem. The key characteristics are defined by the orthogonal arrays and
the signal-to-noise ratio to create a reduced model scale. SVM-RFE is an algorithm
created by SVM-based function selection. The selected key and important feature
sets are used with SVM-RFE. In addition to reducing calculation time, the rating
accuracy rate can be improved. In recent years, many scientists have enabled this
method to improve the classification effect in medical diagnoses.
 An innovative IoT-based breast cancer monitoring system 161

SVM-RFE is a recursive feature elimination program focused on the rating

criteria of weight magnitude. The algorithm in the linear case can be represented in
the following by using SVM-train.

Algorithm of SVM-RFE
Training examples X0 ¼ ½x1 ; x2 ; . . . xk ; . . . xl T
Class labels y ¼ ½y1 ; y2 ; . . . yk . . . yl T
Initialize:
Subset of surviving features s¼[1,2, . . . n]
Feature ranked list r¼[]
Repeat untils¼[]
Restrict examples of training to successful indices
X¼X0(:,s)
Train the classification
a¼SVM-train(X, y)
X Calculate the dimension length(s) weight vector
w¼ ak yk xk
k Calculate parameters of rating
ci¼(wi)2, for all i
Find the function with the lowest classification criteria
f¼argmin(c)
Update feature ranked list
r¼[s( f)r]
Delete the function with the smallest classification criteria
s¼s(1: f1, fþ1: length(s))
Output:
Feature ranked list r.
As described earlier, for speed reasons, more than one function can be removed
by stage.

9.3.4 SMO algorithm

The SMO algorithm is now given in a short summary. The basic step is to select the
index pair (i1, i2) and only change ai1 and ai2 to maximize the dual objective. As
the working set can be used to remove one of the two Lagrange multipliers only in
size 2 and the equivalent restriction [23], the optimization problem in each step is
quadratics in a single variable. A research solution for it is simple to write. All
details need not be remembered at this point. We only create unique significant
point about the function of the parameter threshold. In place of Platt, the pro-
ductivity error in the ith pattern is defined as

Ei ¼ Fi  b

Let us mark the indexes of the two multipliers chosen to optimize in one stage
as i2 and i1 according to the Platt [8] pseudo code. An analysis of Platt’s
162 The Internet of medical things

information (1998) shows that we just need to learn about Ei1  Ei2 ¼ Fi1  Fi2 to
take a step from ai1 and ai2 .
For an efficient solution to the problem, the method used for the choice of i1
and i2 at each stage is essential. Platt developed a strong set of heuristics, based on a
variety of experiments. The two-loop method is used: the external loop is i2, and the
internal loop is i1 for the chosen i2. The external loop iterates over all patterns,
violating the optimal conditions—first only over those with Lagrange multipliers at
either the upper or lower boundary of optimality conditions to ensure that the
problem has in fact been solved. Platt maintains and updates an Ei cache for indices
i for no boundary multipliers for efficient implementation. The rest of the eggs are
calculated as necessary.
Now let us see how i1 is chosen by the SMO algorithm. The goal is to increase the
objective function significantly. Because all the possible i1 options to try to pick one that
best enhances the target function are expensive, the i1 index is chosen to optimize
j Ei2  Ei1 j. Since Ei is available in cache for non-boundary multiplier indices, only the
abovementioned i1 indices are used initially. If the option of i1 does not provide ade-
quate improvement, the following steps are taken. All indexes of no bound multipliers,
one by one, are evaluated as i1 choices from a randomly selected set. If enough progress
is not made, all indices are attempted, one at a time, from a randomly chosen index as i1
options again. The random seed collection, therefore, affects SMO’s running time.

9.3.5 Reforms of the SMO algorithm

Within this section, there are two updated SMO variants, both of which deal with
the problems in the last paragraph. Well beyond the first SMO algorithm, almost all
of the developments are much better and in most cases they are much better
checked. Suppose Fi is open to all at any time i.
Let i_low and i_up be indices s.t.
Filow ¼ blow ¼ maxfFi :i 2 I0 [ I3 [I4 g
Fi up ¼ bup ¼ minfFi :i 2 I0 [ I1 [I2 g

Then it is easy to test a unique i for optimality. Suppose i 2 I1 [ I2 for instance.

We just need to test if Fi<Fi_low2r. If that is the case, the SMO takeStep process can
also be applied to the index pair (i, i_low). For indices in the other sets, similar steps can
be taken. Thus, as opposed to the original SMO algorithm, our solution involves
searching for optimal i2 and selecting the second index i1 in hand. We will see later that
through an efficient updating process, we calculate and use (i_low, blow) and (i_up, bup).
1. We would like to be successful by using a great deal of the energy, as with the
SMO algorithm ai 2 I0 , to modify the Fi, i 2 I0 cache. Only if optimality is
available, search for all the indicators for optimality.
2. The takeStep protocol [23] involves a few additional steps. When the pair of
indices (i2, i1) is successfully used, let I ¼ I0 [ fi1 ; i2 g. Notice that these extra
steps are inexpensive, since fFi; i 2 I0 g is available and updates of Fi1 ; Fi2 are
simple to perform. A careful look reveals that each of the two sets,
 An innovative IoT-based breast cancer monitoring system 163

I \ ðI0 [ I1 [ I2 Þ and I \ ðI0 [ I3 [ I4 Þ, is nonempty; hence, the partially cal-

culated (i_low, blow) and (i_up, bup) will not be zero elements, because i2 and i1
have only been involved with the effective move. Since in the next stage (see
element 1 earlier) I will take values from i_low and i_up and they will be used
as i1 options, we also cache the Fi1 and Fi2 values.
3. Only if the loop with examine All¼0 is used with /i ; i 2 I0 , it must be noted that
if equation blow  bup þ 2t is maintained at some point, it implies optimality as
far as I0 is involved. (This is because the blow and bup choice is affected by all the
indicators in I0 as stated in item 2.) This makes it easy to leave this circuit.
4. Two ways to enforce the loop with indices only in I0 (examine All¼0) are
provided.
Process 1. This is in line with SMO. Close all i2 2 I0 . Search for optimal
values for each i2 and pick i1 accordingly, when split. For instance, if
there is a breach of ion Fi2 < Flow  2t, choose i1¼i_low in this case.
Process 2. Function always through the poorest violator couple, pick
i2¼i_low and i1¼i_up.
The subsequent cumulative alteration of SMO, SMO—Alteration 1 and SMO—
Modification 2, depends upon which one of these methodologies is used. SMO and
SMO—Modification 1 are the same except for the test of optimality. SMO—
Modification 2 can be viewed as an incremental upgrade to SMO—Modification 1,
by efficiently using the cache to select a violation pair while examine All¼0.
5. When optimality is achieved with I0, all indices (examineAll¼1) are checked
for optimal results. Here, we loop one by one across all indices. I has been
computed only partially using I0 since. (blow, i_low) and (bup,i_up) we update
the quantities each of them as they are examined. Fi is first measured for a
given i and optimum control is rendered using the current (blow, i_low) and
(bup,i_up); Fi is used to change the amounts without any breach. There is an
infringement, for example, when i 2 I1 [ I2 and Fi2 < Flow  2t in which case
we are making changes to the SMO algorithm of Platt.

9.3.6 Breast cancer prediction proposed predictive

framework
The protocols for the new BC program (algorithm) are as follows. Figure 9.1 dis-
plays the flowchart of the device.
Algorithm for predictive method of BC
Start
Stage 1: Information preprocessing of BC using techniques of preprocessing
Stage 2: SVM-REF algorithm for attribute evaluation
Stage 3: Process of training and evaluating divisions for the data division
Stage 4: Prepare the training dataset SMO predictive model
Stage 5: SMO predictive model validation with test dataset
Stage 6: Computes models of assessment performance metrics such as TP rate,
FP rate, precision, recall, MCC, and ROC area.
End
164 The Internet of medical things

Breast cancer
dataset

Preprocessing

Attribute
evaluation

Data partition

Classification

Malignant
Benign

Figure 9.1 Predictive method of breast cancer

Table 9.1 Confusion matrix

BC subject predicated Healthy subject predicated

BC subject predicated TP FN
Healthy subject predicated FP TN

Metrics performance evaluation: Assessment metrics used for evaluating the

classifier’s efficiency. Three methods of performance evaluation have been used in
this study. Table 9.1 displays the binary classification problem uncertainty matrix.
The following metrics and the equations are determined according to Table 9.1.
TP, when classified as BC Subject (true positive)
TN, if a good subject is considered healthy (true negative)
FP, if a healthy subject is classified as BC (false positive)
FN, if the BC is listed as safe (false negative)
Classification Accuracy AC: The precision shows the total classification system
output. Precision is the correct probability of diagnosis executed:
TP þ TN
AC ¼
100%
TP þ TN þ FP þ FN
TP rate: True positive rate (correct classification of instances)
FP rate: False positive rates (fake class instances)
 An innovative IoT-based breast cancer monitoring system 165

Precision: Proportion of cases that truly belong to a class separated by the

whole class:
TP
Precision ¼
100%
TP þ FN

Recall: Proportion of cases classified as a class separated by the actual class

total (equivalent to the rate of TP):
TP
Recall ¼
100%
TP þ FN

MCC: MCC is used to measure the quality of binary classifications (two

classes) in machine learning. It takes real and false positive and negative con-
siderations into account and is generally considered to be a balancing measure that
can be used, even if the groups are very different:
TP
TN  FP
FN
MCC ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
100%
ðTP þ FPÞðTP þ FN ÞðTN þ FPÞðTN þ FN Þ

ROC area: Area calculation of ROC (receiver operating properties): one of

Weka’s major output values. You get an understanding of how the classifiers
function in general.

9.4 Study and discussion of experimental findings

Each section uses an algorithm for functionality selection to perform experiments
for the prediction of BC. For the prediction of BC, the machine learning pre-
dictive model SVM was used. Dr. William Wolberg developed the dataset
“WDBC,” which can be downloaded from the UCI Machine Learning library
[20]. The dataset is now accessible. This dataset is used in mobile and wireless
communication.
For training and testing purpose, the dataset is divided into 70%. Varies per-
formance assessment measures such as TP rate, FP rate, precision, recall, MCC,
and ROC area are used to control the predictive model performance. All perfor-
mance measurements are automatically calculated. Preprocessing methods are used
on a dataset for data enhancement before implementing attribute evaluation algo-
rithms and predictive analysis models. In addition, the tables show all these
experimental findings and some graphics have also been developed to better
explain them. Both tests have been performed on Weka.

9.4.1 Preprocessing for the dataset

Table 9.2 offers details and a summary of 569 instances with 32 dataset features
and some statistical measures that are automatically measured [24]. The class dis-
tribution of the dataset consists of 357 benign and 212 malignant subjects.
Table 9.2 Feature details and definition of certain Wisconsin Diagnostic Breast Cancer statistical indicators

Feature Name of feature Description Min Max Mean Standard deviation

1 id Integer
2 radius_mean Mean of distances from the center to points 6.981 28.110 14.127 3.524
on the perimeter cell
3 texture_mean The standard deviation of grayscale values 9.710 39.280 19.290 4.301
4 perimeter_mean Perimeter of cell 43.790 188.500 91.969 24.299
5 area_mean Area of cell 143.500 2 501.00 654.889 351.914
6 smoothness_mean Local variation in radius lengths 0.053 0.163 0.096 0.014
7 compactness_mean Perimeter2/area—1.0 0.019 0.345 0.104 0.053
8 concavity_mean The severity of concave portions of the contour 0.000 0.427 0.089 0.080
9 concave points_mean Number of concave portions of the contour 0.000 0.201 0.049 0.039
10 symmetry_mean Symmetry 0.106 0.304 0.181 0.027
11 fractal_dimension_mean “Coastline approximation”—1 0.050 0.097 0.063 0.007
12 radius_se 0.112 2.873 0.405 0.277
13 texture_se 0.360 4.885 1.217 0.552
14 perimeter_se 0.757 21.980 2.866 2.022
15 area_se 6.802 542.200 40.337 45.491
16 smoothness_se 0.002 0.031 0.007 0.003
17 compactness_se 0.002 0.135 0.025 0.018
18 concavity_se 0.000 0.396 0.032 0.030
19 concave points_se 0.000 0.053 0.012 0.006
20 symmetry_se 0.008 0.079 0.021 0.008
21 fractal_dimension_se 0.001 0.030 0.004 0.003
22 radius_worst 7.930 36.040 16.269 4.833
23 texture_worst 12.020 49.540 25.677 6.146
24 perimeter_worst 50.410 251.200 107.261 33.603
25 area_worst 185.200 4 254.00 880.583 569.357
26 smoothness_worst 0.071 0.223 0.132 0.023
27 compactness_worst 0.027 1.058 0.254 0.157
28 concavity_worst 0.000 1.252 0.272 0.209
29 concave points_worst 0.000 0.291 0.115 0.066
30 symmetry_worst 0.157 0.664 0.290 0.062
31 fractal_dimension_worst 0.055 0.208 0.084 0.018
32 diagnosis
 An innovative IoT-based breast cancer monitoring system 167

Feature information:
(1) id number
(2–31) Ten real-valued features are computed for each cell nucleus:
(a) radius (mean of distances from center to points on the perimeter)
(b) texture (standard deviation of grayscale values)
(c) perimeter
(d) area
(e) smoothness (local variation in radius lengths)
(f) compactness (perimeter2/area—1.0)
(g) concavity (severity of concave portions of the contour)
(h) concave points (number of concave portions of the contour)
(i) symmetry
(j) fractal dimension (“coastline approximation”—1)
(32) Diagnosis (M¼malignant, B¼benign)

A digitized image of a fine needle aspirate of a breast mass is used to measure

the characteristics. They define features of the cell nuclei that are present in the
picture. Separating plane mentioned earlier was obtained using Multisurface
System-Tree [25], a system of classification that uses linear programming to con-
struct a decision tree. Related features were picked in the space and 1–3 separating
planes using an exhaustive scan. In this section, we compiled ten 3D visualizations
from Figures 9.2 to 9.11, each of which is calculated as X-axis, Y-axis, and Z-axis
for every cell nucleus of mean, se, and worst. In respective pole, this includes the
value from min to max.

9.4.2 Results of SVM-RFE experiment

In this reason, as with SVM-RFE, a risk of potential performance degradation may
eliminate many functional variables per stage for computational efficiency.
Therefore, it is important to normalize the values of each function variable across
the samples in order to use SVM-RFE. Nevertheless, since the selection of func-
tions is a step toward creating a good classification, if a better subset of functions is
chosen, a computer-intensified way would be worthwhile.
For selection purposes, the weights of a linear discriminating classifier may be
used. More specifically, one starts with all genes in a backward selection process and
extracts the most informative gene iteratively. The gene which has the lesser effect
on the costing role of the classification procedure is considered in deciding the fea-
ture to be removed in each iteration. The remaining features of the target mark are the
most significant. Table 9.3 records the findings of the SVM-RFE algorithm.

9.4.3 Results of SMO classification

9.4.3.1 Tests of SMO (Normalized PolyKernel) classification
For prevision of the full range of features and on various ranked feature subset
generated by SVM-RFE algorithms, and tabled in Table 9.3, the SMO
Figure 9.2 Radius (mean of distances from center to points on the perimeter)—
range of X: 6.981–288.110, Y: 0.112–2.873, Z: 7.930–36.040

Figure 9.3 Texture (standard deviation of grayscale values)—range of X:

9.710–39.280, Y: 0.360–4.885, Z: 12.020–49.540
 An innovative IoT-based breast cancer monitoring system 169

Figure 9.4 Perimeter—range of X: 43.790–18.500, Y: 0.757–21.980, Z:

50.410–251.200

Figure 9.5 Area—range of X: 143.500–2 501.000, Y: 6.802–542.200, Z:

185.200–4 254.000
170 The Internet of medical things

Figure 9.6 Smoothness (local variation in radius lengths)—range of X:

0.053–0.163.110, Y: 0.002–0.031, Z: 0.071–0.223

Figure 9.7 Compactness (perimeter2/area—1.0)—range of X: 0.019–0.345, Y:

0.002–0.135, Z: 0.027–1.058
 An innovative IoT-based breast cancer monitoring system 171

Figure 9.8 Concavity (severity of concave portions of the contour)—range of X:

0.000–0.427, Y: 0.000–0.396, Z: 0.000–1.252

Figure 9.9 Concave points (number of concave portions of the contour)—range of

X: 0.000–0.201, Y: 0.000–0.053, Z: 0.000–0.291
172 The Internet of medical things

Figure 9.10 Symmetry—range of X: 0.106–0.304, Y: 0.008–0.079, Z: 0.157–0.664

Figure 9.11 Fractal dimension (“coastline approximation”—1)—range of X:

0.050–0.097, Y: 0.001–0.030, Z: 0.055–0.208
 An innovative IoT-based breast cancer monitoring system 173

Table 9.3 Thirty features and their ranking created by SVM-RFE

Feature Name of feature Min Max Mean Standard

deviation
1 radius_worst 7.93 36.04 16.27 4.83
2 concave points_worst 0.00 0.29 0.11 0.07
3 perimeter_worst 50.41 251.20 107.26 33.60
4 texture_worst 12.02 49.54 25.68 6.15
5 concave points_mean 0.00 0.20 0.05 0.04
6 area_worst 185.20 4 254.00 880.58 569.36
7 symmetry_worst 0.16 0.66 0.29 0.06
8 radius_mean 6.98 28.11 14.13 3.52
9 smoothness_worst 0.07 0.22 0.13 0.02
10 area_mean 143.50 2 501.00 654.89 351.91
11 radius_se 0.11 2.87 0.41 0.28
12 texture_mean 9.71 39.28 19.29 4.30
13 perimeter_mean 43.79 188.50 91.97 24.30
14 concavity_mean 0.00 0.43 0.09 0.08
15 compactness_se 0.00 0.14 0.03 0.02
16 perimeter_se 0.76 21.98 2.87 2.02
17 fractal_dimension_mean 0.05 0.10 0.06 0.01
18 concavity_worst 0.00 1.25 0.27 0.21
19 area_se 6.80 542.20 40.34 45.49
20 symmetry_mean 0.11 0.30 0.18 0.03
21 compactness_mean 0.02 0.35 0.10 0.05
22 smoothness_mean 0.05 0.16 0.10 0.01
23 smoothness_se 0.00 0.03 0.01 0.00
24 fractal_dimension_se 0.00 0.03 0.00 0.00
25 fractal_dimension_worst 0.06 0.21 0.08 0.02
26 texture_se 0.36 4.89 1.22 0.55
27 concavity_se 0.00 0.40 0.03 0.03
28 concave points_se 0.00 0.05 0.01 0.01
29 symmetry_se 0.01 0.08 0.02 0.01
30 compactness_worst 0.03 1.06 0.25 0.16

(kernel¼Normalized PolyKernel) model performance predictive model was tested.

All our experiments use SMO parameters as batchSize¼100, buildCalibration
Models¼false, C¼1.0, logistic calibrator, epsilon¼1.0E12, numDecmal
Places¼2, numFolds¼1, randomSeed¼1, and toleranceParameter¼0.001.
Tables 9.4 and 9.5 record the performance of the SMO Normalized PolyKernel
predictive model on a particular combination of features. SMO’s Normalized
PolyKernel has 95.08% correctly classified with TPRate of 95.1%, FPRate of 6.6%,
precision of 95.1%, recall of 95.1%, 89.4% MCC, and ROC area (plotted in
Figure 9.12) of 94.3%.

9.4.3.2 Tests of SMO (PolyKernel) classification

The SMO model performance predictive model (kernel¼PolyKernel) was tested
to provide the full range of features and identify feature subsets generated by SVM-
174 The Internet of medical things

Table 9.4 Correctly and incorrectly classification accuracy of SMO

Kernel TP FN FP TN Correctly Incorrectly

classified (%) classified (%)
Normalized PolyKernel 193 19 9 348 95.08 4.92
PolyKernel 203 9 1 356 98.24 1.76
PukKernel 206 6 0 357 98.95 1.05
RBFKernel 178 34 1 356 93.85 6.15

Table 9.5 Weighted average accuracy % of SMO classifier

Kernel TPRate FPRate Precision Recall MCC ROC area

Normalized PolyKernel 95.3 6.1 95.3 95.3 89.8 94.6
PolyKernel 98.2 2.8 98.3 98.2 96.3 97.7
PukKernel 98.9 1.8 99.0 98.9 97.8 98.6
RBFKernel 93.8 10.2 94.3 93.8 87.1 91.8

0.5

0
0 0.5 1

Figure 9.12 ROC of Normalized PolyKernel SOM X: false-positive rate Y: true-

positive rate
 An innovative IoT-based breast cancer monitoring system 175

0.5

0
0 0.5 1

Figure 9.13 ROC of Linear Kernel SOM X: false-positive rate Y: true-

positive rate

RFE algorithms and tabled in Table 9.3. All the experiments use SMO parameters
like batchSize¼100, buildCalibrationModels¼false, C¼1.0, epsilon¼1.0E12,
numDecmalPlaces¼2, numFolds¼1, randomSeed¼1, and tolerance para-
meter¼0.001. Tables 9.4 and 9.5 record the output in a specific feature combination
of the SMO polykernel predictive model. The polykernel of SMO has 98.24%
correctly classified with 98.2% TPRate, 2.8% FPRate, 98.3% precision, 98.2%
recall, 96.3% MCC, and ROC area (plotted in Figure 9.13) of 97.7%.

9.4.3.3 Tests of SMO (Puk) classification

Tested to provide full feature sets and define feature subsets developed by SVM-RFE
algorithms and tabled in Table 9.3, the SMO model performance predictive
(kernel¼PukKernel). For any experiment, SMO parameters are used, for example:
batchSize¼100, constructionCalibrationModels¼false, C¼1.0, epsilone¼1.0E12,
numDecmalPlaces¼2, numFolds¼1, RandomSeed¼1, and parameter¼0.001 for
tolerance. The performance of the SMO PukKernel prediction model is shown in
Tables 9.4 and 9.5 in a certain function combination. The SMO PukKernel contains
98.95% correctly classified with 98.9% of TPRate, 1.8% of FPRate, 99.0% of preci-
sion, 97.8% of recall, 98.6% of MCC, and 98.3% of the ROC area (plotted in
Figure 9.14).

9.4.3.4 Tests of SMO (RBF) classification

The SMO model performance predictive (kernel¼RBFKernel) is checked to
provide complete feature sets and set feature subsets generated with SVM-RFE
176 The Internet of medical things

0.5

0
0 0.5 1

Figure 9.14 ROC of PukKernel SOM X: false-positive rate, Y: true-positive rate

0.5

0
0 0.5 1

Figure 9.15 ROC of RBFKernel SOM X: false-positive rate Y: true-positive rate

 An innovative IoT-based breast cancer monitoring system 177

algorithms and tabled in Table 9.3. SMO parameters, for example: batchSize¼100,
constructionCalibrationModels¼false, C¼1.0, epsilone¼1.0E12, numDecmalPlaces¼2,
numFolds¼1, RandomSeed¼1, and parameter¼0.001 tolerance shall be used for
any experiment. A certain function combination is shown in Tables 9.4 and 9.5 to
demonstrate the efficiency of the SMO RBFKernel prediction model. The
RBFKernel SMO includes 93.85% correctly classified with 93.8% TPR, 10.2%
FPR, 94.3% precision, 93.8% retrieval, 887.1% MCC, and 91.8% of the ROC area
(plotted in Figure 9.15).

9.5 Conclusion
The IoT, which has become a means of processing both actual data and past
knowledge by the evolving part of cloud computing and data mining, has seen the
transition to life in recent years. A diagnostic method is built in this chapter ana-
lysis for the diagnosis of BC. SMO was used to detect BC in the design of the
device computer prediction model. For the proper and relevant selection of the
correct target type of malignant and benign persons, the SVM-RFE attribute
assortment algorithm is used. New sets of characteristics from the Diagnostic BC
dataset were developed by the SVM-RFE algorithm. The dataset was intended for
training and validation. In addition for model performance assessment, techniques
for measuring performance such as correctly and incorrectly classification accu-
racy, TP rates, fp rates, precision, retrieval, MCC, and ROC area were also used.
The WDBC dataset was developed to check the proposed program using 32
features with 30 existent importance features and 569 instances on the UC Irvine
data mining depository. Weka machine learning libraries are used to incorporate
and improve the proposed framework. The test results indicate that the method
suggested efficiently classifies the malignant and benign. Various changes to the
BC characteristics may result in an increase in malignant and benign predictions.
These results indicate that the new diagnostic method can be used to assess BC
reliably and to be incorporated into health care in addition. SVM classifications
were evaluated on a rating subset of features using several kernels, such as the
Normalized PolyKernel, linear, Puk, and RBF. SMO PukKernel-linear efficiency
is higher than other SMO-kernels according to Tables 9.4 and 9.5 and Figure 9.16
graph also indicating. The novelty of the study is planned to distinguish BC and
stable people as a diagnostic device. For the diagnosis of BC, the program used
the FS algorithm SVM-RFE, SMO, training methods, and output assessment
methods.
Machine-based decision-making is more effective for improved detection of
BC. According to Table 9.5, the proposed device performance (SVM-RFE-SMO),
compared with the classification results of other studies proposed, is excellent
and achieves 98.9% classification accuracy. More technology will be used in
future to boost the performance of BC diagnostic framework with the addition of
assortment algorithms, optimization, and dedicated neural network classification
approaches.
178 The Internet of medical things

360
340
320
300
280
260
240
220
200
180
160 NormalizedPolyKernel
140
120 PolyKernel
100
80 PukKernel
60
40 RBFKernel
20
0

ea
TP

FP
FN

s TN
as d %

te

Pr te
on

ll

RO C
ca

C
ra

ra

ar
isi
d

M
Re
ly sifie

ie
TP

FP

C
ec
sif
la
cl
c
In ctly

ct
rre

rre
Co

co

Figure 9.16 Enactment appraisal of SMO-altered kernels for BC analysis

Table 9.6 Performance of other classification results %

Year [Reference] Model Results

2017 [16] Ensemble of CNN 96.39
2017 [26] SVM and KNN 98.57
2018 [27] IG-SVM 98.83
2018 [28] PCA-AE-Ada 85
2018 [29] Ensemble adaptive voting 98.5
2018 [30] SAE-SVM 98.25
Proposed SVM-RFE-SMO 98.9

References
[1] American Institute of Cancer Research, “Breast Cancer Statistics,” 2018,
https://www.wcrf.org/dietandcancer/cancertrends/ breast-cancer-statistics.
[2] A. J. Jara, M. A. Zamora, and A. F. G. Skarmeta, An Internet of Things–
Based Personal Device for Diabetes Therapy Management in Ambient
Assisted Living (AAL). Springer, London (2011).
[3] M. Islam, H. Iqbal, R. Haque, and K. Hasan, “Prediction of breast cancer
using support vector machine and K-nearest neighbors,” in Proceedings of
the IEEE Region 10 Humanitarian Technology Conference (R10-HTC),
vol. 23, pp. 1–5, Dhaka, Bangladesh, 2017.
[4] A. M. Ahmad, G. M. Khan, S. A. Mahmud, and J. F. Miller, “Breast cancer
detection using Cartesian genetic programming evolved artificial neural
 An innovative IoT-based breast cancer monitoring system 179

networks,” in Proceedings of the 14th Annual Conference on Genetic and

Evolutionary Computation, pp. 1031–1038, Philadelphia, PA, USA, 2012.
[5] A. T. Azar and S. A. El-Said, “Probabilistic neural network for breast cancer
classification,” Neural Computing and Applications, vol. 23, no. 6,
pp. 1737–1751, 2013.
[6] E. Warner, M. Hans, C. Petrina, E. Andrea, S. Rene, and P. Donald,
“Systematic review: using magnetic resonance imaging to screen women at
high risk for breast cancer,” Annals of Internal Medicine, vol. 148, no. 9,
pp. 671–679, 2008.
[7] S. K. Pal, T. Kesti, M. Maiti, et al., “Geminate charge recombination in
polymer/fullerene bulk heterojunction films and implications for solar cell
function,” Journal of the American Chemical Society, vol. 132, no. 35, pp.
12440–12451, 2010.
[8] J. Platt, “Sequential minimal optimization: A fast algorithm for training sup-
port vector machines.”
[9] E. AliăźKović and A. Subasi, “Breast cancer diagnosis using GA feature
selection and rotation forest,” Neural Computing and Applications, vol. 28,
no. 4, pp. 753–763, 2017.
[10] M. F. Akay, “Support vector machines combined with feature selection for
breast cancer diagnosis,” Expert Systems with Applications, vol. 36, no. 2,
pp. 3240–3247, 2009.
[11] B. Zheng, S. W. Yoon, and S. S. Lam, “Breast cancer diagnosis based on
feature extraction using a hybrid of K-means and support vector machine
algorithms,” Expert Systems with Applications, vol. 41, no. 4, pp. 1476–
1482, 2014.
[12] G. N. Ramadevi, “Importance of feature extraction for classification of
breast cancer datasets, a study,” International Journal of Scientific and
Innovative Mathematical Research, vol. 3, pp. 763–368, 2015.
[13] H. A. Abbass, “An evolutionary artificial neural networks approach for
breast cancer diagnosis,” Artificial Intelligence in Medicine, vol. 25, no. 3,
pp. 265–281, 2002.
[14] A. Onan, “A fuzzy-rough nearest neighbor classifier combined with
consistency-based subset evaluation and instance selection for automated
diagnosis of breast cancer,” Expert Systems with Applications, vol. 42, no.
15, pp. 6844–6852, 2015.
[15] R. Sheikhpour, M. A. Sarram, and R. Sheikhpour, “Particle swarm optimi-
zation for bandwidth determination and feature selection of kernel density
estimation based classifiers in diagnosis of breast cancer,” Applied Soft
Computing, vol. 40, pp. 113–131, 2016.
[16] R. Rasti, M. Teshnehlab, and S. L. Phung, “Breast cancer diagnosis in DCE-
MRI using mixture ensemble of convolutional neural networks,” Pattern
Recognition, vol. 72, no. 24, pp. 381–390, 2017.
[17] R. Ani, S. Krishna, N. Anju, M. S. Aslam, and O. S. Deepa, “IoT based
patient monitoring and diagnostic prediction tool using ensemble classi-
fier,” in Proceedings of the 2017 International Conference on Advances in
180 The Internet of medical things

Computing, Communications and Informatics (ICACCI), Udupi, India,

2017.
[18] Z. Yang, Q. Zhou, L. Lei, K. Zheng, and W. Xiang, “An IoT cloud based
wearable ECG monitoring system for smart healthcare,” Journal of Medical
Systems, vol. 40, p. 286, 2016.
[19] R. Royea, K. J. Buckman, M. Benardis, et al., “An introduction to the
Cyrcadia Breast Monitor: A wearable breast health monitoring device,”
Computer Methods and Programs in Biomedicine, vol. 197, p. 105758,
2020.
[20] W. H. Wolberg, Wisconsin Diagnostic Breast Cancer (WDBC). University
of Wisconsin School of Computer Science, UCI Machine Learning
Repository, Madison, WI, USA (1995).
[21] S. Kotsiantis, “Data preprocessing for supervised learning,” International
Journal of Computer Science, vol. 1, pp. 111–117, 2006.
[22] I. Guyon, J. Weston, S. Barnhill, and V. Vapnik, “Gene selection for cancer
classification using support vector machines,” Machine Learning, vol. 46,
pp. 389–422, 2002.
[23] S. S. Keerthi, S. K. Shevade, C. Bhattacharyya, and K. R. K. Murthy,
“Improvements to Platt’s SMO algorithm for SVM classifier design,” Neural
Computation, vol. 13, no. 3, pp. 637–649, 2001.
[24] M. H. Memon, J. Ping Li, A. Ul Haq, M. H. Memon, and W. Zhou., “Breast
cancer detection in the IoT health environment using modified recursive
feature selection,” Wireless Communications and Mobile Computing,
vol. 2019, p. 19, 2019, Article ID 5176705.
[25] M. M. Islam, H. Iqbal, R. Haque, and K. Hasan, “Prediction of breast cancer
using support vector machine and K-nearest neighbors,” in Proceedings of
the 2017 IEEE Region 10 Humanitarian Technology Conference (R10-
HTC), vol. 10, pp. 226–229, Dhaka, Bangladesh, 2017.
[26] N. Liu, J. Shen, M. Xu, D. Gan, E.-S. Qi, and B. Gao, “Improved cost-
sensitive support vector machine classifier for breast cancer diagnosis,”
Mathematical Problems in Engineering, vol. 2018, p. 13, 2018, Article ID
3875082.
[27] D. Zhang, L. Zou, X. Zhou, and F. He, “Integrating feature selection and
feature extraction methods with deep learning to predict clinical outcome of
breast cancer,” IEEE Access, vol. 6, pp. 28936–28944, 2018.
[28] N. Khuriwal and N. Mishra, “Breast cancer diagnosis using adaptive voting
ensemble machine learning algorithm,” in Proceedings of the 2018 IEEMA
Engineer Infinite Conference (eTechNxT), pp. 1–5, New Delhi, India, 2018.
[29] Y. Xiao, J. Wu, Z. Lin, and X. Zhao, “Breast cancer diagnosis using an
unsupervised feature extraction algorithm based on deep learning,” in
Proceedings of the 2018 37th Chinese Control Conference (CCC), pp. 9428–
9433, Wuhan, China, 2018.
[30] K. P. Bennett, “Decision tree construction via linear programming,” in
Proceedings of the 4th Midwest Artificial Intelligence and Cognitive Science
Society, pp. 97–101, 1992.
 Chapter 10
Patient-centric smart health-care systems for
handling COVID-19 variants and future
pandemics: technological review, research
challenges, and future directions
Adarsh Kumar1, Saurabh Jain1, Keshav Kaushik1 and
Rajalakshmi Krishnamurthi2

Information technology can play a vital role in future smart health-care systems.
Using information technology, health-care services can be improved. This
improvement includes shifting the specialization or department-centric health-care
services to patient-centric health-care services. This shift is necessary to have better
patient experiences and providing specialized health-care services to many patients
with lesser resources. In information technology, the Internet of Things (IoT) is an
advanced approach for ensuring this system. In IoT, industrial IoT (IIoT), Internet
of nano things, Internet of robots, Internet of patients, and Internet of medical
things (IoMT) are some of the concepts important to understand the functionality of
smart health-care system. In addition to IoT or its variants, other IoT-associated
solutions that include blockchain technology, parallel and distributed computing
approaches (cloud/fog/edge), virtualization, cybersecurity, automated software
development, and smart infrastructure development have shown great enhance-
ments in recent times. The objective of this work is to explore different
information-technology-based solutions that can make a patient-centric smart
health-care system feasible in nearby times. In this work, recent developments of
IoT that are used to interconnect health-care objects and made technical revolutions
will be explored initially. Thereafter, IoT association with other technological
approaches will be explored. IoT association with other technical aspects is
necessary to understand the IIoT-based health-care solutions. The survey work will
start with the integration of IoT technology with blockchain technology to keep the
patient data confidential and transparent to authenticated parties. The IoT devices

1
Department of Systemics, School of Computer Science, University of Petroleum and Energy Studies,
Dehradun, India
2
Department of Computer Science and Engineering, Jaypee Institute of Information Technology, Noida,
India
182 The Internet of medical things

integrated with the patient help in collecting the data which will be stored at some
central repository. Here, blockchain maintains data security via cryptography pri-
mitives and protocols. The cryptography primitives will provide confidentiality,
integrity, authentication, access control, and non-repudiation properties, whereas
blockchain technology ensures immutability, transparency, distributed computing
and ledger, and data security. In IoT networks, security is a major concern. IoT
networks contain both resourceful and resource-constraint devices. Both types of
the device-based networks require different security solutions. In any smart health-
care system, both types of device-based solutions are required. This work will
discuss feasible cybersecurity solutions for IoT networks useful in smart health-
care systems that are developed in recent times. The feasible cybersecurity solu-
tions will consider health-care infrastructure as cyberspace for collecting, storing,
and analyzing data. In a similar but larger infrastructure, a vast amount of parallel
and distributed computing approaches are required. In this work, we will be dis-
cussing various approaches that are used in recent times for parallel and distributed
computing in health-care system. This includes cloud, fog, edge, and message
passing interface (MPI)-based approaches. All these approaches are used in recent
times at a large scale for IoT-based applications. These approaches will be required
to understand and compute various healthcare-related tasks using IoT networks and
provide efficient outcomes. Thereafter, data-related technological aspects that
include the usage of artificial intelligence (AI), and machine learning (ML) will be
explored. These aspects will address the recent developments that use AI- and ML-
based approaches and are found to be useful in the smart health-care system. In this
technological aspect, all associated approaches, including deep learning, reinfor-
cement learning, federated learning, neural network, virtual reality (VR), and
augments reality aspects, derived during recent times will be explored.
Virtualization has played an important role in reducing the number of resources and
improving the services. Thus, a few of the recent aspects in the direction that are
associated with smart health-care systems are planned to be discussed here. In a
smart health-care system or Healthcare 4.0-based technological solution, data
privacy and security are other major concerns. The traditional security primitives
and protocols are easily breakable with quantum computers. Post-quantum cryp-
tography aspects are required to handle future security aspects. Thus, this will
discuss the importance of quantum computations for IoT networks. Further, how
post-quantum cryptography will be helpful, recent developments for the health-care
system and applications to automated patient-centric services will be discussed.
Lastly, robotics and drone-based technological solutions that are recently discussed
in the scientific community and important for Healthcare 4.0 or smart health-care
system will be explored. This study will briefly explain the recent medical opera-
tions and experimentations that are conducted successfully using robots and drones.
All the information-technology-based solutions require smart infrastructure. Thus,
requirements of smart infrastructure in the smart health-care system which include
smart electric supply system, smartphone-based mobile applications, smart medical
and pharmaceutical system, smart ambulance and other transportation systems, and
smart medical appliances will be explained.
 Patient-centric smart health-care systems 183

10.1 Introduction
Software systems with advanced technologies, including AI, ML, drones, cognitive
computing, parallel and distributed processing, smart and intelligence processing,
and interconnection of devices, have played important roles in various applications
with different forms of their implementations. There are technological aspects like
Healthcare 4.0, Industry 4.0, Agriculture 4.0, and Agriculture 5.0 that integrated
major technological aspects under one umbrella for automation. Thus, the form and
technological aspects vary from application to application. In a smart and intelli-
gent health-care system, these technical aspects have played and will continuously
play vital roles. The major advantaged that information-technology-based aspects
can give to various applications include (i) health-care automation; (ii) patient-
centric health-care services deployment rather than specialization or department-
centric approaches; (iii) remote monitored, control, and analysis-based effective
system; (iv) health-care datacenter creation for effective analysis and handling
futuristic situations; (v) industrial health-care system with large-scale inter-
connectivity; and (vi) secure and trustworthy health-care services. With these
aspects taken into consideration, the major and important functionalities of
healthcare-related applications are discussed in recent times concerning fast and
secure processing approaches toward automation. The effective and efficient
approaches can only be important to implement if a detailed analysis shows its
prediction for a real-time system. The use of AI and ML approaches has made this
possible with the help of advanced resources. Thus, there is a strong need to discuss
and explore such recent advancements in the medical and health-care sector.
This work has surveyed the recent developments in the health-care sector for
handling medical services to patients. In recent developments, technological
aspects are covered in detail. Among technological aspects, those approaches are
found which can handle the COVID-19 or its related situations effectively. The
survey of the proposed system includes the interconnection of medical devices for
patient-centric services. It has been observed that most existing health-care services
are specialization-based services with the least interconnectivity. Lack of con-
nectivity and automation make the medical services slower for patients. This
increases the overall cost and time for health-care operations as well. Thus, there is
a strong need to shift from a specialization-based health-care system to a patient-
centric health-care system. In recent times, various initiatives have been taken
using advanced technologies. This work discussed these technological systems
briefly.
This work is organized as follows. Section 10.2 covers the healthcare-related
advancements using IoT-based technology. Similarly, Section 10.3 covers the need
and advancement of blockchain technological aspects for health-care systems and
services. The interconnection of medical systems, patients, equipment, and pro-
cesses needs data security. Section 10.4 covers health-care cyberspace and asso-
ciated cybersecurity aspects. These aspects are necessary to handle a patient-related
health-care system. Large-scale implementation of health-care services is not pos-
sible without the use of parallel and distributed computing architectures. Thus,
184 The Internet of medical things

important, and recently proposed parallel and distributed computing approaches for
healthcare are discussed in Section 10.5. Section 10.6 discussed the recently pro-
posed AI and machine-learning-based approaches for effectively handling health-
care services and pandemic situations. Section 10.7 discusses the importance of
virtualization for the effective utilization of resources in handling smart health-care
systems. In the future, the quantum computational aspect is necessary for handling
large health-care data and fast processing. Thus, Section 10.8 discusses the
importance of quantum computation and its role in a smart health-care system.
Post-quantum computing and cryptography is the solution to those problems that
identify loopholes in quantum-based breakable solutions. In the health-care system,
data privacy and security is very important. Thus, Section 10.9 discusses the role of
post-quantum computing and cryptography to effectively handle futuristic pan-
demic situations. Section 10.10 discusses the recently implemented and/or pro-
posed drone and robot-based approaches for medical and smart health-care
services. Finally, the conclusion and futuristic aspects are drawn in Section 10.11.

10.2 Internet of Things (IoT) network for patient-centric

health-care system

In recent COVID-19 timings, it has been observed that IoT networks played
important roles in handling patients. Many wearable devices can help everyone to
monitor their health-related activities and plan the future. Singh et al. [1] identified
the integration of wearable devices and IoT networks to patient-centric smart
health-care systems. It has been identified that the proposed wearable and IoT-
network-based system can reduce the chances of medical mistakes that include
doctor, medicine, and operational mistakes as well. Further, the proposed system is
capable of handling patients with lesser expenses, efficient machinery for superior
treatment, effective medical control, and enhanced medical diagnosis. An IoT-
based system for handling medical situations (like COVID-19), the important
processes include (i) health data monitoring in remote location, (ii) virtual man-
agement of meetings and conferences with sensor- and camera-based devices, (iii)
data collection, storage, analysis, and visualization for effective control and mon-
itoring, and (iv) case report preparation and presentation for effective display and
handling at various levels. Kamal et al. [2] have discussed the importance of the
patient health-care system in COVID-19 handling, its present situation, challenges,
and opportunities. This is a small comprehensive survey explaining the systems of
integrating wearable and sensor devices attached to a patient’s body. With this
integration, patient monitoring is much easier and automated. Likewise, the
Internet of cameras can be constituted which constantly monitor the patient and
hospital activities. In this monitoring, image-based analysis is possible which can
constantly monitor patient movements, emergencies, remote monitoring, and sur-
veillance. In similar cases, a camera-based interconnected system is helpful to
handle hospital situations. The administration and surveillance are much easier and
can be done with minimum manpower.
 Patient-centric smart health-care systems 185

10.2.1 IoT and its applications for health-care sector

This section discusses the applications of IoT or its variants to smart health-care
systems and services. There are various applications of IoT networks to health-care
systems. Some of the popular applications are briefly explained as follows:
1. Sensor-based biodegradable chips or implantable sensors can be attached to a
patient’s skin that continuously gives glucose monitoring and reading to the
patient or his/her doctor over the phone. This way, it is very easy to self-
monitor the regulation of sugar level in the human body.
2. In activity-tracking-based applications, IoT sensors can help in adjusting the
dietary and exercise plans as per his movements, habits, fatigue, etc. Thus, IoT is
helpful to health-care professionals in effectively handling their patients with
well-accepted data. With this feasibility, it is much easier to control overall
health plans rather than focusing on overweight and age-based calculations only.
3. Like blood sugar measurements, wearable IoT devices can be used for high
blood pressure measurements as well. In this experimentation, the device will
continuously monitor the variations in blood pressure. These measurements
when made accessible to patients or health-care service providers, then it would
be easy for them to monitor heart-related activities. For example, strokes, pal-
pitations, and full-brown heart attacks. A well-connected system can arrange
medical, ambulance, or doctor services on time whenever the chances of severity
increase. Thus, the IoMT made it possible to monitor patients at a large scale and
provide specialized services as and when they are required.
4. A family-interconnected IoT system can help adults to monitor their family
members and their health conditions. For example, if there are old people in the
family and it is necessary to remotely monitor their health, an IoT-based system
can help them a lot. As discussed earlier, they can monitor old people by looking
at their health-related data variations, the system can give auto-generated signals,
or a well-connected camera-based IoT network can send real-time images. With
the help of these images, it is much easier to monitor patients and provide them
with necessary services as and when it is required with remote accessibility only.
5. IoT-network with the sensor-based system can help the patient to monitor their
medicine dosages as well. It is much easier to check the quality and quality of
medicines. Besides, analysis such as time of medicine effects of medicine over
the body, right medication, etc. can be monitored using infectible sensors. In
this way, it is much easier for any patient to handle their disease well on time
rather it causes severe effects.
6. The IoT-based monitoring system can help the patients or their family members
to note the timings of their medicines and refill as and when it is necessary.
7. The IoT-based system helps health-care service providers to monitor their
equipment, medicines, and infrastructure-related status continuously with partial
to full automation. This way, it is possible to reduce the efforts of manpower and
increases the feasibility of everything work well on time and on a large scale.
8. The IoT-based health-care system is effective for all stakeholders, including
patients, doctors, hospitals, health-care service providers, and insurance
186 The Internet of medical things

companies. This is because it creates a well-connected and trustworthy envir-

onment to share reliable information on time. The availability of this infor-
mation ensures that it can be securely stored at the right locations and can be
analyzed for timely treatment and diagnosis of diseases.
9. Among other advantages, IoT-based health-care services provide cost reduc-
tion in health-care systems and improving medical operations and related ser-
vices after analyzing reliable data.
Table 10.1 shows the comparative analysis of open-source IoT applications for
health-care system proposed in recent times.

Table 10.1 A comparative analysis of open-source IoT application for health-

care system

Source System Description

[3] IoT-healthcare for end-to-end con- In this implementation, a healthcare-related system
nectivity, simulation, and report- is designed for reporting and monitoring. Using
ing this system, end-to-end connectivity is possible.
Additionally, item tracking, alert generation,
remote medical assistance, and research are
performed to understanding IoT-based health-
care system. It has been observed that data
privacy and security, system integration, over-
head handling and cost are important parameters
to design an effective health-care system
[4] Raspberry Pi and Arduino-based In the proposed system, raspberry pi and Arduino-
patient’s data reading system board-based patient data reading system is
proposed for real-time health data collection
from sensors. The proposed system sends
collected data to the doctor and notifies the
system. This sharing of information helps
patients and doctors to continuously monitor
patient’s readings
[5] IoT-based health-care system to The proposed system collects heart rate data from
collect heart rate 85 subjects. The data is collected with different
features from subjects taken into consideration.
The proposed system detects future heart rates
using a random forest model
[6] AWS-based IoT model for health- The proposed health-care system integrated IoT
care services options with cloud services for managing
notifications. Here, notifications are immedi-
ately sent over to digital devices whenever a
vital report is observed. The implemented push–
pop model is effective in handling medical data
sharing between different stakeholders
[7] Health statistics application using The proposed health-care application is sponsored
JavaScript by Philips Healthcare. This application is
developed for patient data collection and
sharing with doctors. In this process, data is
made continuously with web display and it is
easy to understand and diagnose the patient
 Patient-centric smart health-care systems 187

10.2.2 Industrial IoT (IIoT) for smart health-care system

This section discusses a few of the recent and important developments in extending IoT
networks to IIoT. In IIoT, sensor, wireless, and application-specific objects can be
interconnected at a large scale for industrial applications. This connectivity is faster
than response generation, data sharing, and service facility to a large scale. Some of the
recent developments in IIoT and its health-care applications are discussed as follows.
Mumtaz et al. [8] discuss the importance of 5G and beyond networks for IIOT
growth. In this discussion, the application-specific industrial level is on focus for vast
and rapid industrial-level development. The health-care sector is one such major
target that is achievable through IIoT. Using IIoT, health-care services can be
extended to a large scale at a comparatively lesser cost. A chain of hospitals inte-
grated with the IIoT network can focus its service on the patient-centric health-care
system rather than specialization-based services. This way, it would be easier for
patients to get his treatment under one umbrella. Further, it is easier for the health-
care service provider or hospitals to provide their services as well. Rathee et al. [9]
discussed the importance of integrating IIoT with blockchain technology. In this
work, an IIoT framework is discussed and the importance of integrating blockchain
technology with this framework is explored. It is recommended that if health-care
services are required to be extended with IIoT network, there is a strong need to
integrate blockchain technology as well because it can bring transparency, immut-
ability, distributed computing, ledger, and cryptography-integrated security. All of
these things are important for a patient-centric health-care system. In a patient-centric
health-care system, if records are permanently stored, it would be much easier for any
health-care service provider or doctor to see historical data. This data can be made
available to any location in a secure way. Thus, this framework is good for those
medical services, services of which are not easily available or they are rare.
10.2.3 Resourceful and resource constraint device-based
smart health-care developments
IoT devices in the health-care sector can be resourceful or resource-constraint. The
resourceful devices like laptops, desktops, display screens, cameras, and other large
computational and storage-capable devices have an abundance of resources for com-
puting cryptography primitives and protocols. On the other hand, resource constraint
devices, including Radio Frequency Identification (RFID) tag, RFID reader, and sen-
sors, have a scarcity of resources that cannot implement traditional cryptography pri-
mitives and protocols. To ensure security over these devices, lightweight cryptography
primitives and protocols are required. Here, lightweight means lesser hardware
resources in terms of XOR, OR, NAND, NOT, etc. gates. The RFID and sensor
devices are useful in health-care applications in different scenarios. Medicines,
equipment, machines, and patients can be identified with RFID easily.
10.2.4 Research challenges and future direction of IoT in
patient-centric smart health-care system
In literature [7–9], it has been observed that there are many challenges to IIoT
networks. For example, the Quality of Service (QoS) in the IIoT network is a major
188 The Internet of medical things

challenge for adopting it to large-scale application-specific usages. In QoS,

throughput, delay, jitter, connecting topology, the number of secure connections,
and more factors influence the IIoT network. Besides, developing patient-centric
health-care services is not easier in real scenarios because of the lack of medical
infrastructure in a large number of developing or underdeveloped countries. To
handle these challenges, there is a need to integrate various information-
technology-based recent developments into the health-care sector. For example,
integrating AI and machine-learning-based technology in the health-care sector
helps in faster data analysis. Thus, if a large amount of data is generated from every
device of IoT or IIoT network, analyzing this much of data is difficult. The best
solution is to apply ML, deep learning, AI, and cognitive computing aspects for
faster and efficient analysis of results. Likewise, the IIoT network can be integrated
with 5G and beyond network technologies for fast data accessibility and response
generation. Availability of advanced Internet technologies ensures the availability
of data as and when it is required. Thus, providing high QoS-based technologies for
data sharing is important, which makes the real-time data available to health-care
insurance, doctor, and health-care service providers. Similarly, blockchain and
cryptography primitives can ensure the security of patient’s data. Patient’s data
security and privacy is a major concern in the health-care sector. Ensuring these
services can build trust in this system. To handle a large amount of data generated
in IoT or IIoT networks, parallel and distributed computing approaches like cloud,
fog, edge, and MPI-based computing should be integrated. To make faster the
computational at an individual end, quantum computing can be used. This type of
computing can faster the storage, transmission, and processing responses. To secure
the data against quantum computing hacks, post-quantum cryptography mechan-
isms can be helpful. Thus, future challenges are (i) hoe to integrated IoT networks
with advanced technologies such as blockchain, AI/ML, quantum computing, par-
allel and distributed computing, and post-quantum methodologies. Among all,
patient data security and privacy are a prime concern. To achieve this target,
cryptography primitives and protocols can be integrated at various levels. In tra-
ditional computing infrastructure, both resourceful and resource-constraint devices
are available. Thus, traditional and lightweight cryptography solutions can be
implemented. However, the advanced quantum-computing-based infrastructure
required fast responding cryptography primitives and protocols. To handle tradi-
tional cryptography primitives and protocols against quantum computing attacks,
post-quantum cryptography can be used.

10.3 Blockchain technology and IoT for patient-centric

health-care system

There has been a rapid increase in health-care data in recent years since medical
care has become the most common and necessary part of human lives. To make
medical analysis and prescription comfortable and suitable for patients as well as
doctors, the health sciences are now making their way toward the IoT. The
 Patient-centric smart health-care systems 189

implementation of the IoT in healthcare has made remote monitoring easy. Patients
can now be monitored remotely without the need to visit hospitals periodically. All
the data is shared directly with doctors, and the prescription can be made accord-
ingly. But this personal data of patients needs protection as sharing and storing data
using IoT is not secure and can be hacked easily. So, implementation of the
blockchain in smart health-care systems can provide security to the data of the
patients [10,11].

10.3.1 Need for the blockchain and IoT integration

Blockchain technology is one of the most emerging technologies nowadays.
Blockchain is used almost in every field starting from healthcare, finance, educa-
tion, and whatnot. The decentralized and autonomous nature of blockchain tech-
nology makes it suitable for integration with IoT networks. The blockchain-based
approach can provide tangible solutions to various privacy and security issues in
IoT. The integration of smart contracts into blockchain technology has made it
possible to handle various possibilities in achieving reliable decentralization. The
following is a list of features that make blockchain a unique technology [12,13].
● Security: In blockchain, data from IoT devices is stored as a transaction record
in a block. Before data is stored in a block, each transaction is mined and
validated by all other peer users using a consensus approach inside the
blockchain. This guarantees information security.
● Decentralization: Decentralization means that there the power is not residing
in a single authority which means that there is no such central authority. In a
decentralized system, a group of authorities is held responsible for handling
everything making it decentralized.
● Scalability: Scalability is the outcome of decentralization which improves
fault tolerance.
● Immutability: Immutability means that once changes are done, they cannot be
altered or changed after the changes are done making a blockchain an unal-
tered network. Immutability makes a blockchain corruption-free hence con-
sidered as a key feature of this.
● Transparency: Blockchain promises users regarding transparency and providing
a pseudonymous system. Blockchain makes everything transparent making peo-
ple argue that blockchain could be used as the new standard for transparency.

10.3.2 Role of Internet of Things in health-care system

HealthBank [14], a global Swiss digital health start-up, offers an approach to the
transaction and sharing of personal health data. This start-up provides its users with
a platform on which they can keep and manage their health records in a protected
environment. Users can fully hold data sovereignty in their hands. As a next step,
the HealthBank start-up applies blockchain technology to its underlying business
model. With the help of blockchain technology, patients can get their health
information such as eating habits, blood pressure, diabetes information, sleep
190 The Internet of medical things

patterns, and heart rate transparently and safely. Patients can use the health app or
ask for information from doctors or hospitals. Today, using the HealthBank
approach, not only patients are able to place their information on this platform, but
it is also available for medical research purposes. This innovation has expanded a
new direction of vulnerability and security for data from IoT-based networks.
Implementation of traditional cryptographic solutions on IoT networks is not a
future aspect as IoT devices have many constraints such as performance, memory,
and computational power. In this work [15], the authors have proposed a novel
structure, a unique architecture that integrates blockchain networks with IoT
devices to achieve privacy and security threats to information integrity. Integration
of smart contracts is very important in this work, by which we can handle author-
ization, access control, data management, and device authentication. The overall
scalability of the network can be improved using an off-chain data storage
mechanism in the framework, this model can be integrated into any existing IoT
application with minimal modifications.

10.4 Cybersecurity and IoT for patient-centric health-

care system
With the advancement in technology, interconnected smart devices are evolving.
The domain of the IoT is giving various opportunities in the domain of healthcare.
Various IoT-based health-care devices are assisting patients and doctors across the
globe. These IoT-enabled smart devices are boon to the various medical practi-
tioners and patients but on the other hand, if the security of such devices is not
implemented properly, it can create some serious trouble. The growth of the IoT in
healthcare is affecting people’s lives. For people suffering from such illnesses, the
use of new and advanced wearable devices makes life easier. With such functions,
these wearable devices are produced that make them an appropriate portion of the
human body. To provide the desired performance, wearables are also assisted by
recent technologies such as edge computing. The role of wearables in IoT health-
care, as well as the reference architecture of new wearable devices and various IoT
communication protocols, is highlighted in [16]. Smart devices interpret data from
their surroundings, make calculations on it, and then either save or transmit the data
to a computer with similar capabilities. In this communication process, there are
many security and trust concerns, which are really important to manage. These
security and trust issues will pose potential challenges to the IoT environment, and
the current architecture, framework, and technology need to be tackled. Several
important security and confidence risks in the IoT and how they are handled by the
new architecture are seen in [17].

10.4.1 IoT and cybersecurity

Advances of medical care provision innovations are increasingly proliferating
worldwide. In a plethora of new rules, policies, and legislation, these developments
are being tackled piecemeal. Currently, these efforts frequently struggle to put the
 Patient-centric smart health-care systems 191

patient and his or her well-being at the center of changes in legislation, policy, and
regulations. Rather, attempts are being made to balance patient results, financial
benefits, and liability concerns. Increasingly, an increasingly acute danger to
patients is the failure to recognize and handle cybersecurity problems associated
with their own well-being. In [18], the argument for patient-centered approaches to
not only medical treatment but also health-care cybersecurity is discussed. The
development of the IoT is still in its infancy and it is important to solve several
similar problems. The IoT is a coherent notion of combining all. IoT has a tre-
mendous opportunity to make accessibility, honesty, availability, scalability, anon-
ymity, and interoperability more available to the public. How to secure the IoT,
however, is a difficult challenge. Device security forms the basis for IoT growth. In
[19], IoT reviews cybersecurity. Protecting and incorporating heterogeneous smart
devices and information technology systems are essential concerns (Information and
Communications Technology). The architectural framework in relation to cyberse-
curity in IoT comprises various features and principles, some protocols related to the
architecture, synchronization and wireless networking, heterogeneous and pervasive
frameworks, lightweight technologies, authentication, etc. The architecture involves,
from a technological perspective, integrity, scalability, reliability, secrecy, avail-
ability, and interoperability between heterogeneous smart devices.
Numerous risks are emerging in the IoT paradigm, including unauthorized or
inappropriate operation, malicious code updates, control bypassing, and data integrity
tampering. In IoT implementations, information may be revealed or lost.
Consequently, safeguarding sensitive documents, passwords, and certificates is cru-
cial. Provided that, unregulated IoT apps and systems reveal embedded patented
algorithms that can easily be pirated or studied, intellectual property can be breached.
To stop the discovery of hidden bugs, it is advised that hackers find it more difficult
to reverse-engineer, study, or manipulate the code in general. To ensure continuity
and authenticity of data, integrity processes are used. Hash functions and digital
signatures are used to ensure the confidentiality of documents. In addition, at the
storage level as well as on the network path in the IoT environment, data protection
must be secured. The anonymity program is the concealment of sources of data. In
addition, this program helps in the preservation of personal privacy and secrecy. In
the IoT, non-repudiation [20] helps to guarantee that a bargaining side is unable to
challenge on official papers the authenticity of its signature. Finally, freshness sig-
nifies the presence of facts and the lack of past texts. Figure 10.1 shows the inte-
gration of IoT and cybersecurity and related factors necessary to understand.
Device or network protection is a key issue, and best practices such as restricting
external computer connections, restricting the usage of the Internet directly to some
important endpoints and devices, ensuring that only designated networks are author-
ized, using keys for safe booting and safe firmware are used in numerous ways to
secure them, enforcing device authentication at each link establishment. Virtual
Private Networks are external networks that allow only partners to access them, which
they pledge to, keep private and have assured honesty. This infrastructure, on the other
hand, is not visible for dynamic global information sharing and is not safe for non-
extranet third parties. DNS Security Extensions use asymmetric cryptography for asset
192 The Internet of medical things

Linking IoT and cybersecurity

Supporting Implemention
Possible threats Specifications points
technologies

Anonymous Risky
Confidentiallty Applicability
routing operations

Virtual private Networks and

Data theft Integrity
networks gateways

DNS Reverse
Availability System or
extensions for
engineering services
defence

Private Intellectual Non-

recovery of Utilities
details property theft repudiation

Figure 10.1 Linking IoT and cybersecurity

record signing to ensure the root validity and integrity of information received or
delivered. Data is wrapped in multiple layers of encryption using the public keys of
onion routers on the transmission network, for example, using onion routing tech-
nology to encrypt and mix the Internet traffic of several senders.

10.4.2 Recent developments in ensuring confidentiality,

integrity, and availability (CIA) properties in IoT
networks for smart health-care system
For those who require regular and real-time medical monitoring and therapeutic
procedures, the integration of IoT with medical devices leads to promoting the
consistency of health-care facilities and progress data on patient status. IoT accel-
erates early illness identification and facilitates the prevention and recovery pro-
cess, such as exercise services, infectious illnesses, and caring for the elderly. With
all the gains, IoT adoption is followed by the possibility of new security breaches
and health-care system flaws. The following explanations are associated with this:
(i) medical devices primarily capture and exchange confidential patient informa-
tion, (ii) the design of IoT technologies causes problems of sophistication and
incompatibility, and (iii) medical IoT system vendors do not pay attention to
security features. Due to the abovementioned factors, confidentiality, integrity, and
availability security [21] concerns are increasing. The apps and technologies that
track and manage the vital signs of patients are some of the IoT solutions in
healthcare. These technologies can, however, be subjected to security threats, such
as authentication, permission, and privacy breaches. Cyber protection in the world
of healthcare has been a major concern. Hackers can take the advantage of com-
puter vulnerabilities and contribute to IoT system operational disturbance. More
 Patient-centric smart health-care systems 193

specifically, due to the limitations of medical equipment, including power usage,

scalability, and interoperability, conventional security requirements for attack coun-
termeasures are not applicable. Not only can unauthorized access to IoT components
have detrimental implications for the running of processes, but also it more fre-
quently presents a risk to human life and health. The intent of intruders may be to
destroy some industrial facilities, copy records, intercept vital equipment systems
maintenance, and other acts aimed at preventing the proper functioning of equip-
ment. The inability of developers [22] to use simple techniques and methods of
securing knowledge has resulted in various flaws and allegations of “hacking” IoT-
solutions. Second, it should be remembered that suppliers at the design level of the
systems do not take the classification of the data being handled into consideration. As
a result, the established instruments deal with a wide variety of classified details,
including personal information, business secrets, clinical secrets, and so on. The bulk
of the bugs, on the other hand, could be removed by using safe data transfer methods.
Many cloud links remain vulnerable to attack using transport-level encryption
mechanisms like Secure Sockets Layer/Transport Layer Security.
IoT is basically a network of interconnected entities. Obviously, even though
they have access to it, it is inexpedient to accept a group of objects that do not
communicate within the network. Consequently, the data-sharing mechanism
between objects is crucial. IoT enterprises also turn to automated message
screening and review of transmitted data while embedding multiple technologies.
Much of the time, these features are provided to make interacting with the device
smoother, such as configuring incoming updates. Any of them, on the other hand,
are intended to screen out potentially dangerous or faulty data transmitted from
computers to humans. Personal medical devices patients will stay mobile and
healthy for even longer, thus preventing the need for supportive housing. This type
of gadget normally has a wireless interface that allows it to connect with a base
station that can read medical data, monitor system status, change settings, or update
the computer’s firmware. This wireless software introduces the user to security
dangers, raising the patient’s security and privacy issues. The study also discovered
that this system and its wireless interface are vulnerable to cyberattacks, placing
healthcare, defense, and privacy at risk. Research teams have shown a number of
attacks on medical devices, including snooping, message manipulation for security
and reputation, and battery draining attacks to cause availability issues. Various
countermeasures are proposed for these threats. When designing countermeasures
for them, the cost constraints on medical devices for implementing protective
measures are a major concern. Battery capacity is the most important consideration
for implantable medical devices [23]. The batteries in computers are typically non-
rechargeable and must be replaced every few years.

10.4.3 IoT cyberspace and data handling processes for

information and network security
Because of their wide scope, openness, and comparatively high processing speed,
IoT artifacts have been an ideal target for cyberattacks. Furthermore, since many
194 The Internet of medical things

IoT nodes store and process private data, they are turning into a gold mine for
cybercriminals. As a consequence, security, especially the ability to detect com-
promised nodes, as well as the ability to catch and preserve evidence of an attack or
malicious activity, emerges as a top priority for effective IoT network deployment.
In the IoT world [24], anonymity, access control, safe networking, and secure data
storage are also becoming significant security issues. Furthermore, during an
inquiry, every new device we develop, every new sensor we deploy, and every byte
that is organized in an IoT environment can be scrutinized. The rapid growth of IoT
devices and software has resulted in the introduction of various unstable and
insecure nodes. Furthermore, traditional user-driven protection architectures are
inefficient in object-driven IoT networks. Therefore, to protect IoT networks,
obtain, store, and analyze residual information from IoT settings, we need
advanced equipment, techniques, and procedures. Health programs and resources
are more important, innovative, and linked than ever before. While enhancing
clinical quality and changing patient delivery, thus improving human life, protec-
tion of health-care data and devices is becoming a growing concern. Health
equipment and services have grown more intertwined, opening them to emerging
cybersecurity risks. It makes the health-care industry the most vulnerable to serious
safety threats. As mentioned earlier, the problem is compounded by the health-care
services and infrastructures allowed by Cyber-Physical Systems (CPS)-IoT [25] that
are vulnerable to a number of emerging cyberattacks. CPS-IoT systems are cate-
gorized as essential safety and protection systems and present extended attack surface
characteristics of fragmentation, interconnectedness, heterogeneity, and cross-
organizational existence. Cybersecurity threats have the ability to breach users’
privacy, inflict bodily harm, financial losses, and pose a danger to human life, so
avoiding them is vital. With millions of medical documents [26] stolen worldwide,
studies illustrate the increase of attacks and the rise of medical identity fraud.

10.4.4 Research challenges and future direction of IoT in

patient centric smart health-care system
The IoT movement involves linking a huge number of smartphones and wireless
gadgets that are traditionally placed at the bottom of the Internet spectrum and
capable of delivering large amounts of data at a high rate. The Internet data flow on
its back has the potential to alter this trend. It is simply impractical or impossible to
gather all of the data and send it to datacenters present remotely then wait for the
reports to be transmitted back to the edge after large data processing. Essentially,
two variables would preclude it. The first aspect [27] is the amount and speed at
which the IoT devices collect the generated data. Data collection and transmission
on a local basis is unavoidable. The second aspect is that many potential IoT
technologies such as autonomous vehicles, VR, augmented reality, and involving
AI have a low latency or rapid response requirement to make quicker decisions. In
terms of technological developments, intelligent devices with high communication
and data collection capacities are now possible, opening up a slew of new oppor-
tunities for IoT applications, particularly in healthcare. There are also a range of
 Patient-centric smart health-care systems 195

open topics with many of the benefits that discuss big IoT concerns, such as
usability, portability, interconnection, data storage, and privacy. IoT provides a
systemic basis for a range of high-tech health-care technologies, including real-
time patient tracking, environmental and interior safety data, and ubiquitous and
widespread access to information that helps both health professionals and patients
while also providing essential features for health-care networks like usability,
mobility, and extensibility. Continuous scientific advances allow IoT devices to be
created by countless sensing, data fusion, and logging facilities, leading to many
advances in better living environments (ELEs) [28]. To transfer data between
sensors and servers, smart healthcare relies on a combination of short- and long-
range communications networks. Wi-Fi, wireless metropolitan area network
(WiMAX), Zig-Bee, and are the most popular Bluetooth short-range networking
networks, which are mostly used for short smart health-care communications such as
body area network. In smart healthcare, a series of technology such as mobile com-
munication (global system for mobile communication) and General Packet Radio
Service, Long-Term Evolution (LTE) Advanced, and LTE are used to relay data
from a local server to a ground station. Security requires smart healthcare. Even
though smart healthcare relies on the Internet to connect various devices, security is a
major concern. Complex security mechanisms and algorithms are difficult to
implement because of the restricted presence of IoT computers (limited processing
and battery life). As a consequence, 70% of IoT applications [29], including intelli-
gent health-care equipment, are vulnerable to cyberattacks. Consequently, there are
regular threats and security and privacy concerns. From hospital to patient-centric
healthcare, there is an increasing sense of urgency. IoT is designed to be a powerful
enabler by ensuring that computers and cloud storage, as well as acting agents such
as patients, hospitals, research laboratories, and emergency services, are all in har-
mony. Given the benefits of IoT eHealth [30], a variety of challenges must be
resolved. Data storage, scalability, interoperability, device–network–human inter-
faces, usability, and privacy are among others. In terms of variety, length, and
velocity, we can see that knowledge is getting more dynamic. Uniformity will allow
for interoperability between different devices and records. All people, from children
to the aged, should be able to use interfaces that are clear and genuine. Safe infra-
structure, network protocols, less vulnerable to threats in the cloud, and adequate
training for people handling secure data are all required to increase the security of
devices, networks, clouds, and agents.

10.5 Parallel and distributed computing architecture

using IoT network for patient-centric health-care
system

This section discusses the parallel and distributed computing-based approaches pro-
posed for the health-care system in recent times. Here, those frameworks are studied
and explored in detail which is recent and found to be effective for handling pan-
demic situations. Details of recently proposed frameworks are discussed as follows.
196 The Internet of medical things

10.5.1 Cloud computing architectures using IoT network for

health-care system
Featherstone et al. [31] have surveyed the need for technologies in the health-care
system and handling pandemic situations. In this survey, it has been identified that
cloud computing and the web 2.0 framework are found to be very effective in
educational institutions during COVID-19 pandemic times. Likewise, cloud-based
infrastructure can be used for handling future pandemic situations as well. Cloud
services can provide various services, including computation, storage, processing,
and security. Nowadays a large number of applications are developed using cloud
computing for different scenarios. In all of the scenarios, the performances are
found to be better compared to old and present traditional computational approa-
ches. Desai [32] discussed the need for software systems and Bigtech in pandemic
situations. A large number of software systems are in usage for health-care ser-
vices. All of these systems generate big data. This big data analysis is important for
the health-care system. With the usage of cloud computing, it is much easier to
analyze the data and evaluate the statistics. Cloud services are also very helpful in
handling a large number of patients. In patient handling, their contract and histor-
ical data are required to be maintained which is possible easily using cloud ser-
vices. This service adds remote accessibility and fast response time. All of these
services make it possible for software-based health-care systems to handle COVID-
19 situations as well. The software systems are found to categorize the patients
easily and isolate them. This way it is possible to take care of them and their
medical requirements more carefully. With the addition of cloud computing ser-
vices, it can be implemented at an industrial scale. Thus, handling pandemic
situations would be much easier. Gupta et al. [33] have developed communities for
effectively handling the COVID-19 situation. This is a different approach where
smart and intelligence solutions are proposed to effectively handle COVID-
19 situations. In this chapter, a detailed survey is performed to explain the need for
such community development. Kumar et al. [34] have proposed a drone-based
solution to effectively handle the COVID-19 pandemic situation. In this approach,
thermal imaging is applied to collect the patient’s data. This data helps in provide
sanitization and medication services. All of these services use the cloud, fog, and
edge computing aspects. To implement these computing services, different envir-
onments are created that made the COVID-19 handling and integration with the
health-care system possible. Table 10.2 shows the other parallel and distributed
computing approaches for smart and intelligent health-care system proposed in
recent times.

10.6 Artificial intelligence and machine learning

approaches in IoT for smart health-care system
Artificial intelligence and ML are integral parts of any system nowadays. The
important features of this technology include data analysis with the least error and
fast processing, learn from the data and predict future trends, the ability to generate
 Patient-centric smart health-care systems 197

Table 10.2 Comparative analysis of parallel and distributed computing

approaches for smart and intelligent health-care systems

Author Year A B C D E F Remarks

Featherstone 2012 N Y N Y Y N This is a survey article to identify the
et al. [31] current pandemic situation and future
aspects. This article discussed var-
ious challenges in medical and IT-
based systems useful for health-care
services
Desai [32] 2020 N Y N Y Y Y This article discusses the need for
software and system to handle
COVID-19 and pandemic systems.
The designed systems can be useful
for future pandemic handling and
treatments. Using this software, it is
much easier to classify the
pandemic-affected patients and their
handling
Gupta et al. [33] 2021 N Y N Y Y Y This article addresses the development
of communities and handling of
COVID-19 pandemic situations
using a community-based approach.
In a community-based approach,
similar features or functionality-
based groups can be performed to
handle the current or future pan-
demic scenarios
Kumar et al. 2021 N Y N Y Y Y This article addressed the usage of
[34] drones and robotic systems for med-
ical treatment. Here, a major con-
centration is drawn toward the use of
contactless approaches that are safe
for medical treatments. Flying mul-
tiple drones in a geographical region
without collision and maximum en-
ergy utilization for health-care ser-
vice is the prime concern of the
proposed system

A: Patient centric, B: specialization or department centric, C: large-scale deployment using Cloud/Edge/

Fog/MPI, D: helpful in data collection, processing, analysis, or visualization, E: able to handle COVID-
19 situation, F: able to handle futuristic pandemic situations.

or support data visualization for interpretation, and many more are important for
any system. In health-care systems, AI and ML have played vital roles and it is
expected to continue as well. This section will explore the recent advancement of
the use of AI and associated technologies for COVID-19 and its variants.
Figure 10.2 shows the important technical aspects that will be explored in detail.
 Al for robotics and
drone-based system
to effectively
implement
Expert system for health services IoT system with Al for healthcare
healthcare automation and industrial level
implementation

Artificial intelligence

Deep/machine/federated Medical data and clinical data

learning experimentation for activity analysis
data analysis

Al-based approach to
improve healthcare
service life cycle

Figure 10.2 AI-based technological aspects for the smart health-care system
 Patient-centric smart health-care systems 199

All of these technical aspects are associated with AI-based developments. In brief,
the use of AI-integrated technological aspects is explained as follows.
● AI technology with robotics is used to perform various experimentations in the
sanitization, thermal imaging, monitoring, and surveillance. All of these
activities are contactless and important to handle pandemic situations. In past
pandemic situations, these activities helped to identify the patients with
COVID-19 symptoms, sanitization, and other healthcare-related activities. A
large number of developments in developed and developing countries are
made to address the challenges of handling COVID-19-related situation
handling using drones. Likewise, robotics is equally important. In the past
decade, efforts are made to integrate human intelligence behavior to machines
using robotics for performing health-care operations.
● IoT system generates a lot of data as well. The best way to handle this data is
using AI and ML. The classification or categorization of data, its interpreta-
tions, and visualizations are easily possible using ML approaches.
● Medical data analysis can be performed in various sections, including clinical
trials, patient records, operations success rate, and susceptible, infectious,
recovery model. This analysis and their observations are important to respec-
tive sectors. All of this creates more important when there is a need to speed up
the clinical trials and obtain the results in comparatively lesser time duration.
For example, it has been observed that to successfully obtain COVID-
19 medicine, the clinical trials are performed at a much fast rate.
● In healthcare, the patient service life cycle starts from symptoms identification to post-
discharge treatments. All of these aspects need constant monitoring and inspection that
is not possible using manual processing. Automated processes with the use of tech-
nology can make this feasible. In this aspect, AI technology is very useful.

10.6.1 Healthcare, COVID-19, and future pandemic-related

datasets for smart health-care systems
In this section, important COVID-19 and pandemic-related datasets are identified
and explored. Table 10.3 shows the various recently prepared datasets. These
datasets are compared using various parameters, including COVID-19 image
dataset, COVID-19 text dataset, COVID-19 audio dataset, other health-care data-
sets, other pandemic datasets, medicine dataset, and patient-related. Results show
that a large number of studies are performed prepared in recent times to study the
impacts of COVID-19 in society. Most of these datasets are based on patient’s
symptoms (cough, fever, headache, etc.) or image-based scanning. However, there
is a strong need to study the hidden factors in COVID-19 variants and future pan-
demic identification or its chances.

10.6.2 Artificial intelligence in smart health-care system

Technology like AI provides the ability to interpret complex medical data, analyze
images, video, audio, and text records, and help the doctor to treat the patient in
200 The Internet of medical things

Table 10.3 Comparative analysis of datasets available for healthcare, COVID-19,

and pandemics

Dataset A B C D E F G Remark
[35] Y N N N N N N This is an image-based dataset to study
the impact of COVID-19 in recent
times. Here, a comparative analysis
of COVID-19 and the non-COVID-
19 dataset is made to have a better
analysis. This dataset consists of
X-ray and CT chest images
[36] Y N N N N N N This is a dataset of COVID-19 X-ray
dataset. This dataset is divided into
training and test datasets. This data-
set is useful for COVID-19 virus
detection, preliminary diagnosis, and
contributes toward virus control
progress
[37] Y N N N N N N This dataset is image-based one in
which images are classified into six
categories: viral, bacterial, fungal,
lipoid, aspiration, and unknown.
Each of these categories includes
further classification. For example,
viral images are classified as
COVID-19 (SARS-CoV-2), SARS
(SARS-CoV-1), MERS-CoV, Vari-
cella, Influenza, and Herpes. Like-
wise, other categories are classified
[38] Y N N N N N N This dataset contains augmented X-ray
images. In this dataset, images are
classified as all augmented images
and COVID-10 X-ray-augmented
images
[39] Y N N N N N N This dataset contains CT images of a
large number of COVID-19 patients.
This dataset has the classification of
images with COVID-19 and non-
COVID-19 patients. These images
are used by Tongji Hospital, Wuhan,
China for diagnosis and treatment of
patients
[40] Y Y N Y N N N This dataset contains X-ray images
with detailed symptoms. Some of the
symptoms are used to calculate the
COVID-19 score. This analysis is
performed to identify the chances of
COVID-19. Dataset is divided into
three major categories: training,
validation, and testing. Using these
categories, the whole dataset is
(Continues)
 Patient-centric smart health-care systems 201

Table 10.3 (Continued)

Dataset A B C D E F G Remark
divided into two parts. The first part
contains information about those
images that are X-ray images for
distribution. On the other hand, the
second part contains those X-ray
images that are collected to classify
patient distribution
[41] Y Y N Y N N N This is another dataset over COVID-19
and non-COVID-19 symptoms. This
dataset is regularly updated and the
last update was performed on Janu-
ary 28, 2021. This dataset also con-
tains CXR images having positive
and negative COVID-19 cases. Here,
pneumonia is the most common
feature considered for analysis that
can indicate COVID-19 as well.
Here, chest radiography is applied to
identify COVID-19 cases with dif-
ferent patients. This image-based
analysis helps in identifying critical
factors necessary to understand
COVID-19
[42] Y Y N Y N N N This is another image-based dataset that
contains categories like left lung,
right lung, cardiomediastinum, air-
ways, ground-glass opacities, conso-
lidation, pleural effusion,
pneumothorax, endotracheal tube,
central venous line, monitoring
probes, nasogastric tube, chest tube,
and tubings. Images of all these
categories are used in experimenta-
tion to identify the COVID-19 symp-
toms
[43] Y N N N N N N This is another image-based dataset that
contains lung images for identifying
COVID-19-affected patients
[44] Y Y N Y N N N This image-based dataset focuses on
multiple symptoms for identifying
COVID-19 cases. Here, the dataset is
divided into normal patients’ images,
COVID-19-affected images, pneu-
monia symptom images, lung opacity
images, and metadata of all cate-
gories. Thus, this data is very useful
in COVID-9 case identification and
treatments
(Continues)
202 The Internet of medical things

Table 10.3 (Continued)

Dataset A B C D E F G Remark
[45] Y N N N N N N This is another sort dataset with
COVID-19 symptom identification
using comparative analysis of pneu-
monia and normal patient
[46] Y N N N N N N This dataset contains CT images that
can be more than 40 COVID-19
patient’s data for analysis and study
[47] Y N N N N N N This dataset is not publically available
for study
[48] Y N N N N N N This dataset contains COVID-19 fea-
ture-based images that were used in
past for identifying COVID-19 pa-
tients with observable features
[49] Y Y N Y N N N This is an image-based dataset that
classifies the COVID-19 symptoms
into various categories, including
common, less common, and rare. In
the common category, symptoms
include fever, cough, anosmia, fati-
gue, sputum, and shortness of breath.
Among less common features,
myalgia/arthralgia, headache, sore
throat, chills, Pleuritic pain, and
diarrhea are taken for identification.
Likewise, nausea, vomiting, abdom-
inal pain, GU bleeding, nasal con-
gestion, palpitation chest tightness,
hemoptysis, and stroke are taken in
the rare category. In conclusion, this
data is very useful for image and
feature-based analysis
[50] Y N N N N N N This image-based dataset is labeled
with COVID-19 CT scans with left,
and right lungs, and infections.
Dataset is prepared to have a com-
parative analysis of COVID-19 and
non-COVID-19 symptoms with lung
disease features
[51] Y N N N N N N This dataset is prepared with lung and
infection segmentation with limited
annotations, segment COVID-19 CT
scans from non-COVID-19 scans,
and both COVID-19 and non-COV-
ID-19 scans. The annotation and
labeled dataset is useful for easy
classification
[52] Y N N N N N N This is ultrasound-based dataset for
COVID-19 patient identification.
(Continues)
 Patient-centric smart health-care systems 203

Table 10.3 (Continued)

Dataset A B C D E F G Remark
The dataset is helpful for noninva-
sive, cheap, portable and available
for almost all medical facilities
compared to other challenging and
time-consuming options for COVID-
19 identification
[53] Y N N N N N N This is a chest CT-scan-based dataset
for COVID-19 identification and
adds a provision to apply artificial
intelligence in radiology for data
analysis based on advanced innova-
tive technologies
[54] Y N N N N N N This is another CT scan image-based
public dataset available for SARS-
CoV-2 CT scan. This dataset con-
tains 1,252 images for SARS-CoV-2-
infected COVID-19 patients and
1,230 non-SARS-CoV-2-infected
COVID-19 patients. All of this data
is collected from Sao Paulo, Brazil
hospitals. This dataset can be used
for COVID-19 identification and
comparative feature analysis with
non-COVID-19 patient data
[55] Y N N N N N N BIMCV-COVID-19þ is a large dataset
with chest X-ray and CXR (CR,
DX), and computed tomography
images. In this image-based analysis,
COVID-19 patients can be identified
with radiographic findings, patholo-
gies experimentation, polymerase
chain reaction (PCR) prediction,
immunoglobulin (IgG), and immu-
noglobulin M (IgM) diagnosis. In
addition to this, various other test
data is available for analysis. In this
dataset, high-resolution images are
stored with the medical imaging data
structure that includes semantic seg-
mentation of radiographic findings.
In addition to images, patient’s
demographic evolution, method of
projection, and acquisition are also
available for studies
[56] Y N N N N N N This dataset is labeled chest X-ray and
CT-images-based dataset for
COVID-19 identification. This data-
set is weakly labeled. This dataset is
(Continues)
204 The Internet of medical things

Table 10.3 (Continued)

Dataset A B C D E F G Remark
a collection of medical images and
can be used for comparative analysis
of clinical symptoms and clinical
findings of COVID-19 and other
influenza demonstrations
[57] Y N N N N N N This dataset is not publicly available
but is used with AI and machine-
learning-based systems for identify-
ing COVID-19 patients
[58] Y N N N N N N In this work, deep learning is applied
for an image-based dataset for iden-
tifying COVID-19 patients

A: COVID-19 image dataset, B: COVID-19 text dataset, C: COVID-19 audio dataset, D: other health-
care datasets, E: other pandemic datasets, F: medicine dataset, G: patient dataset.

many efficient ways. The use of technology can help in reducing the medical
treatment costs that are increasing day by day. The costly medical treatments are
difficult to select for many especially people living below the poverty line. In the
case of scarcity of resources, and higher medical costs, an AI-based system can
help the patient to self-monitor themselves and reduce the cost to a certain extent.
AI-technology can behave like human intelligence, pattern recognition, data ana-
lysis, anomaly, and outlier detection and prediction. These features can help the
medical system to handle more complex challenges that a human can handle. AI
technology is much rich in handling the problem and finding the solutions. For
example, AI-technology can widely be used for analyzing the COVID-19 symp-
toms in recent times, accelerate the scientific discovery processing to handle
COVID-19, its variants, and future pandemics. Usage of AI technology is not just
helpful for medical staff but it helps administrative staff as well. To administrative
bodies, this system can automate the administrative jobs, accelerate and amplify the
system transparency to staff roles and responsibilities, and remove fabricated ser-
vices. In the medical system, data is very important. To handle the data efficiently,
error detection and removal methods need to be focused upon. This error can occur
in data collected for diagnosis, prognosis, and therapy. In medical diagnosis, patient
data collection, and interpretation using doctor’s knowledge and experience for-
mulate it to a proper diagnosis and therapeutic plans for a patient. In the present
system, all of these plans are prepared by a physician. However, it is expected that
this system can be automated with AI in nearby times. This automation can help the
patients to give their inputs and get the diagnosis in output. An expert system is a
well-known example of this type of treatment. An expert system’s inference engine
can handle complex problems with multiple If-then type rules and transform the
input to actionable outputs. Thus, expert systems are very much useful to doctors in
 Patient-centric smart health-care systems 205

clinical decision-making. To handle COVID-19 patients, infectious disease

knowledge can be represented in the form of production rules. The conditional
statements in production rules can help to emphasize overdiagnosis and manage-
ment of COVID-19-infected patients.

10.6.3 Machine learning aspects in smart health-care

system
Sujath et al. [59] applied the linear regression algorithm to identify the ailment and
pace of COVID-19 cases in India. This analysis has experimented with over Kaggle
dataset and multilayered perceptron and vector autoregression is used for this
experimentation. In the analysis, it has been observed that the main symptoms of
COVID-19 are cold, cough, and fever. COVID-19 can cause respiratory ailment
with manifestations as well. In this experimentation, definition and data combina-
tion is kept persistently for analysis. Figure 10.3 shows an example of a machine-
and deep-learning-based system that can be used for COVID-19 diagnosis and
treatment. This system is picked from the real-life observations and it is divided
into three layers (patient-level, hospital-level, and national level analyses). The
roles of ML at different layers are briefly explained as follows:
● National-level ML usages are important for performing randomized clinical
trials and observations. To analyze the success of new medicines, clinical trials
are performed in a randomized fashion over a different set of populations.
Randomized experiments in medical observation, patient monitoring, medical
services, and other healthcare-related systems generate a large amount of data.
This data analysis can easily be performed through the machine and deep
learning models. In health-care systems, ML models take input from EMR and
image-based medical data and give clinical trials recommendation as output.

Patient diagnosis and Patient level analysis

treatment Pre-disease diagnosis and treatment

Serologic testing PCR testing Rapid in-clinic

antigen testing
Machine and deep learning

Hospital services Hospital level analysis

Post-disease diagnosis and treatment
models

Medical services Non-medical services

Nation level analysis

Randomized clinical trials Policy or decision level diagnosis

Figure 10.3 A machine- and deep-learning-based COVID-19 treatment system

206 The Internet of medical things

Over the clinical set of data, natural language processing can be applied for
generating more medical data that can be further analyzed. This is an iterative
cycle that operates until the desired output is achieved.
● Hospital-level analysis includes hospital services (medical and nonmedical)
that can also generate a large amount of data for analysis. The medical services
include the type of treatment, medicines, operations, equipment, and other
medical support. Among nonmedical operations, administration, payment,
insurance, patient monitoring, and other duties are counted. To automate the
complete medical setup, both medical and nonmedical services have to be
precise that is possible with the least error-based ML approach. For example,
the medicine supplier system should have a record of how, which, and what
medicines are delivered to different hospitals. Analysis of this data can indicate
the future requirements, hospital practices in experimenting with the medicines
and their success ratio, a list of authentic medicine suppliers and distributors in
a specific geographical region, and many more.
● In the patient-level analysis, patient diagnosis and treatments can be analyzed
for a depth study. In this analysis, patient test records that include serology,
PCR, and rapid in-clinic antigen tests data are recorded for COVID-19. This
analysis if analyzed through machine- and deep learning aspects then it is very
much useful for patients to get timely diagnosis and treatment.

10.7 Virtualization and IoT in IoT for smart health-care

system

Virtualization technology is important in the future because of the increase in the

number of working people and advancements in technology. Using virtualization,
there is no need for a patient to visit a doctor instead physician or doctor can attend
to the patient through an online video call. Likewise, the experimentations can be
extended to a large scale for the virtual hospital. Thus, virtualization can help to
handle a large number of patients and it is also useful in providing specialized
doctor services on a large scale. The concept of virtual healthcare is surfacing the
market due to the increasing number of working people and advancements in
technology. Virtual healthcare can be defined as virtual visits to physicians through
video calls, chats, or phones. And this is all possible because of the audio and video
connectivity enabling virtual meeting from anywhere. Sometimes a patient is away
from the city, or unable to reach the clinic due to various surrounding factors. In
these cases, videoconferencing with an off-site health-care practitioner is the most
convenient option.

10.7.1 Operating system and storage virtualization for

desktop- and mobile-based smart health-care
applications
Ambigai et al. [60] discussed the smart health-care application to improve the
quality of life of urban people. The medical records of patients stored securely in
 Patient-centric smart health-care systems 207

case they for virtual machine migration. Smart health-care applications store
patient’s medical records, analyze these records, and monitor the patient’s condi-
tion regularly. Integration of cloud services made the availability of patient records
easy, fast, and to a longer distance. This work has used the Ant colony optimization
algorithm to improve the virtual machine selection for generating a sequence of
random optimized solutions. Here, a large number of iterations are performed that
execute the medical record processing at a much faster rate. Virtual machine
migration using particle swarm optimization and virtual machine selection algo-
rithm minimized the number of virtual machines to be migrated and reduces energy
consumption. The objective of smart health-care application development and
integration with virtualization is to minimize the number of virtual machine
requirements to be migrated, reduce energy consumption, and minimize response
time by handling a large number of users. Demirkan [61] has proposed a framework
to include three-layer architecture for a smart health-care system. In the proposed
framework, three-layer architecture is having an option of the service-oriented
framework to include virtualization of resources. The conceptualized data have
driven model in mobile and cloud-enabled smart health-care system with virtuali-
zation give better performance compared to other approaches. The health-care
organization provides cost-effective solutions to health-care services with IT setup
costs and reduced risks. Figure 10.4 shows the smart health-care system blocks with
virtualization. In this system, inputs can be provided through health-care service
providers with different specializations. These specializations include services to
handle patients, patient data, equipment suppliers, and other services like medicines
distribution and staff arrangements. In an IT-based smart health-care system, ser-
vices operate with different data, applications, and software. In service-oriented
service provider
Health-care

Health-care service provider

Health-care processes and workflow

Service-oriented architecture
data, application and software services
service consumer
Health-care

Virtualization software, resources,

application, and assets

Data information storage and

retrieval system

Storage virtualization and other

historical data storage system

Figure 10.4 A smart health-care system with virtualization

208 The Internet of medical things

architecture, data, application, and software can be divided into two categories:
software services and components. In software services, all technical aspects rela-
ted to individual software execution can be explored, whereas components include
the integration of one or more software services to achieve health-care objectives.
For example, patient treatment requires a historical data retrieval system and a
patient monitoring system. In another example, data may be required to share with
multiple systems. This feature can be achieved through storage virtualization. In
storage virtualization, medical records can be transferred easily and at a higher rate.
Further, if application virtualization is shared with storage, the advantage of a
complete system can be taken. A smart health-care system with virtualization can
give various advantages. Some of the advantages are explained as follows:
● Big-data-enabled intelligence and acknowledgment management systems and
services are very useful to the health-care system because the health-care
sector can collect, analyze, share, and visualize the medical structured or
unstructured data for patient treatment and actionable decisions.
● The patient-centric or electronic-record-centric health-care system is useful for
the future in keeping the data secure and safe. This data can be utilized for
better medical treatments and patient monitoring. This data can be used for
preparing patient profiles and can be interrelated for symptom-related experi-
mentation in medical treatment.
● Data linked with ambulance and other medical services can help the patients to
arrive and leave the hospital as per his/her needs. Availability of such data can
automate the patient arrival and discharge system.
● Availability of data can reduce medical costs, shorten patient stays, and can
implement medical treatments much easily. All this is possible when medical
staff is educated with advanced technology to handle the medical treatments
and can see and understand the variations in medical data.
In the area of integrating virtualization technology with healthcare, very few
initiatives have been taken. This area can be explored with the objective of (i)
identifying the direct and indirect risks associated with the collection, storage,
sharing, processing, and analysis of data at various levels; (ii) apply ML and AI
principles in prior identification of medical operations with the highest potential of
improving the patient’s condition; (iii) to understand the challenges in enhancing
the successful medical practices to the industry or large-scale implementations; (iv)
focus on research in smart health-care systems that offer opportunities to study the
different environments and propose models, frameworks, and methods to improve
and automate the health-care services.

10.8 IoT and quantum computing for smart health-care

system
This section explores the IoT and quantum computing operations integration for the
smart health-care system. Abd EL-Latif et al. [62] explained the importance of IoT
 Patient-centric smart health-care systems 209

and quantum computing integration and proposed a two-phase cryptosystem for

security against quantum computing. Quantum computation and quantum algo-
rithms developed in recent times are mathematically very strong compared to tra-
ditional cryptography primitives and protocols. Thus, quantum walks is the term
used for computational models incorporating quantum algorithms and crypto-
graphy. In [63], a quantum-walks-based privacy-preserving mechanism is proposed
for IoT used in the health-care system. The proposed system is a two-phase system
in which substitution and permutations are computed using quantum walks. These
computations provide security and safety. The simulation results show that there is
enough evidence to prove the robustness of image encryption and experimentation
done for quantum walks. These experiments are conducted for protecting patient’s
data and ensuring privacy. Fernández-Caramés [63] identified the threats to tradi-
tional cryptography primitives and protocols. Quantum computing is important for
cryptography due to its ability in solving key distribution problem and high security
in insecure communication channels. Quantum computing provides high security to
email exchange, financial transactions, digital signed documents, medical data, and
military secret information exchanges. Traditional cryptography mechanisms such
as Rivest, Shamir, Adleman (RSA), Elliptic Curve Cryptography (ECC), Hyper
ECC, ElGamal, and Diffie–Hellman algorithms are integral part of network-based
communication. These algorithms are in use over both resourceful and resource-
constraint devices [65,66]. However, long-term usage against quantum computing
challenges is not feasible for resource constraint IoT devices. In [67], quantum
computing survey is performed. Thereafter, need of post-quantum cryptosystem is
discussed. The need of IoT architecture and challenges are discussed with future
trends to secure both resourceful and resource-constraint devices and networks. The
survey article provides a wide range of survey over post-quantum cryptography,
IoT secure, need to integrated quantum computing with traditional cryptography
system for security analysis and important instruction to handle the future network
against quantum computing challenges.

10.8.1 Introduction to IoT and quantum computing

Quantum technology with low power IoT devices is a major challenge for
designing and developing a secure IoT network having security against quantum
attacks. Quantum technology has various associated terminologies, including (i)
quantum networks; (ii) quantum simulators; (iii) post-quantum cryptography; (iv)
quantum sensors and particle generators atomic clocks; (v) quantum cloud com-
puting; (vi) quantum memories, quantum repeaters, quantum chips; (vii) quantum
software; (viii) quantum computing, and quantum annealers; (ix) quantum materi-
als; and (x) quantum key distribution. All of these terminologies are important for
quantum computations.

10.8.2 Quantum computing and smart health-care system

Smart health-care system requires fast and easy processing results. Among these
results, quantum computing can play important roles, including (i) ensuring
210 The Internet of medical things

security to end devices. Patient’s data security is prime concern in futuristic health-
care systems. Quantum-computation-based security system in smart healthcare can
ensure this at a minimum cost. (ii) Analyzing large data with minimum time
duration. Health-care system needs patient’s data to be analyzed at much faster rate
compared to traditional systems. This analysis can predict the disease at an early
stage and helps in patient-centric treatments as well. Quantum computing plays
transformative effects in many ways. For example, quantum computing can help in
developing supersonic drug design at much faster rate. In pharmaceutical industry,
pharmaceutical developments are lengthy and costly. Here, IT-technologies like
artificial intelligence, human organs-on chip, and silico trials play important roles.
For example, the treatment in the case of Ebola virus, and developing new drugs with
smart algorithm. According to [64], the important use cases of quantum computing in
healthcare include the following: (i) it is helpful in diagnostic assistance. Quantum
computing can diagnose the patients at early stage with more accuracy and effi-
ciency. This is possible because of the use of advanced technology and fast com-
putations; (ii) it is helpful in people healthy with personalized interventions and
precision medicine; and (iii) the careful analysis and processing can optimize insur-
ance premiums and price. Quantum computing can play important role in health-care
sector by analyzing the medical images, detecting edges in images and image
matching algorithms. Thus, quantum computing is very helpful in fast image analysis
and image-aided diagnostics. Using quantum computing, it is much easier to com-
bine multiple datasets, perform analysis, and give results at much higher rate.
Quantum support vector machine is very helpful in enhancing the classification and
performing diagnosis that classify cancerous cells from normal easily.

10.8.3 Challenges of quantum computing to health-care

data and data security
The major challenges of quantum computing architecture in health-care domain
include (i) its feasibility in real scenario for applications like healthcare. So far,
very few quantum computing machines are in practice. Most of the quantum
computing implementations are mathematical in present scenario. (ii) Quantum
computers are required to have interdependencies and correlations among con-
tributions that optimize treatment effectiveness. (iii) To provide precision centric
health status and intervention, transition from umbrella treatment to patient-centric
precision health status and interventions are necessary to be focused. (iv)
Optimizing pricing is another major challenge for quantum computing. In quantum
computation, risks to patients, their health conditions, risk model analysis, and
importance of pricing models are important to understand and realize.

10.9 Post-quantum cryptography solutions for futuristic

security in smart health-care system
This section explores the usage of post-quantum in the health-care sector. To secure
the cyberattacks against quantum computing, various post-quantum approaches are
 Patient-centric smart health-care systems 211

proposed. In literature [67–73], various post-quantum cryptography-based approa-

ches are proposed to secure IoT network that is useful to the health-care system. In
[67], post-quantum cryptography aspects are discussed for IoT networks that can
secure health-care applications and data. This work has discussed the limitation and
future applications of quantum cryptography in the health-care system. It has been
realized that quantum computation can revolutionize the health-care system and
ways of handling patients. Quantum computations are capable of providing the
processing power to forecast futuristic health-care malfunctioning. The existing
health-care industry in a developed or developing country is storing data securely.
However, the biggest threat to this industry is the breach of information which is
possible through quantum computing. To protect the system from quantum attacks,
post-quantum cryptography can play an important role. In healthcare, an attacker
can eavesdrop, modify, spoof, and intercept the information. Post-quantum cryp-
tography can be classified into various forms. These forms and their importance to
the health-care sector are explained as follows.

10.9.1 Code-based cryptography for health-care data and

system
In [71–74], cryptography approaches suitable for health-care systems are discussed.
However, most of the cryptography approaches use traditional encryption/decryption
and authentication algorithms for security. However, code-based cryptography can be
an effective approach for securing a smart health-care system. In code-based crypto-
graphy, the use of a code-based encoding and decoding system is useful. Numerical
and polynomial-based encoding and decoding procedures can make the breaching job
tough for attackers. Thus, this area needs exploration. Further, integration of already
developed code-based cryptography mechanisms with health-care systems needs
experiment. The chances of higher security are expected as the post-quantum crypto-
graphy mechanisms are difficult to break because of their complexity.

10.9.2 Lattice-based cryptography for health-care data and

system
In [75,76], lattice-based cryptography is explored for the health-care system. There
are various lattice-based cryptography mechanisms like encryption/decryption
(NTRUEncrypt, GGH encryption, Peikert’s Ring), the hash function (SWIFFT, LASH),
homomorphic encryption, and signature-based approaches (GGH, NTRUSign,
Learning with Errors). All of these approaches are helpful in health-care systems.
For example, NTRUEncrypt [77] is used for application and data-level security in
healthcare. The use of encryption/decryption is explored but other lattice-based
cryptography is yet needed to experiment for health-care services.

10.9.3 Hash-based cryptography for health-care data and

system
In [67,75–78], the importance of post-quantum cryptography mechanisms for
health-care data and application is studied. In these studies, quantum computation
212 The Internet of medical things

is tried to be integrated with IoT networks and blockchain for providing healthcare-
related services. Quantum computations integrated with blockchain technology
need the implementation of hash computations. In a blockchain network, the hash
function provides security to the network using consensus algorithms. The concept
of Hash-cash is also relevant for integrating immutability and transparency to
transactions.

10.9.4 Multivariate cryptography for health-care data and

system
In [79], the importance of multivariate post-quantum cryptography mechanism is
discussed for different applications, including healthcare. In this work, a proxy
signature mechanism to delegate signature to other is discussed. The requirement of
ensuring secure delegation is used of authentic servers. The use of proxy signature
approaches protects against various quantum adversaries. The multivariate public-
key cryptography, its heuristic proofs, and signature approaches are explored. The
authors have implemented the proxy signature scheme and results are analyzed. In
observations, it has been found that distributed approach using multivariate cryp-
tography can give security to various potential applications including healthcare.

10.9.5 Supersingular cryptography for health-care data and

system
In supersingular cryptography, the use of discrete mathematics adds potential
challenges to attackers. For example, Menezes, Okamoto, and Vanstone approach
for supersingular curves provides higher security compared to traditional crypto-
system. The proposed approach provides a significant improvement in performance
(e.g., bandwidth) and higher security. The use of discrete mathematics in a super-
singular cryptosystem increases the complexity and security that can be useful for
medical data as well.

10.9.6 Post-quantum cryptography application and

research challenges for smart health-care system
The major challenges of integrating post-quantum cryptography aspects with the
smart health-care system are as follows [80–82]: (i) implementation issues that are
impossible to handle in the present scenario. A large number of post-quantum
aspects are theoretical and the lack of sufficient platforms to test in real scenario is
not possible in nearby times. (ii) Post-quantum cryptography aspects are considered
to be more complex than can cause major disruption to systems and stakeholders as
compared to prior migrations. (iii) The transition from existing security infra-
structure to post-quantum cryptography-based infrastructure will be costly and can
cause more disruptive changeover. (iv) Existing security infrastructure is developed
over a long time in consideration with challenges evolved. The major challenge is
the orderly transition from existing application-based scenarios to advanced
applications (like smart healthcare)-based application scenarios. This transition
also expects an orderly follow-up that does not look feasible in nearby times.
 Patient-centric smart health-care systems 213

10.10 Drone and robotics operation management using

IoT network for smart health-care system

This section explores the usage of drone networks in health-care operations.

Recently, various drone and robot-based applications are proposed for COVID-19
operations [83–86]. Among these applications, drone-based medication, sanitiza-
tion, surveillance, and monitoring are popularly used. The use of robots in the
health-care sector is largely observed for surgical operations and hospitality. The
research in drone networks and robotics for healthcare is not new. Over the last two
decades [34,87–90], various advancements are observed in this sector that can be
used in the health-care sector effectively. The recent usages of drones and robotics
are explained as follows.

10.10.1 Medical robots and recent developments

This section explores the usage of robots in the medical and health-care domain. In
recent times, various advancements are made in this sector. Some of the recent
developments are discussed as follows.
Lum et al. [87] proposed an IoT-network-integrated surgical drone. Here,
medical equipment, kits, and medicines are remotely arranged with minimum
human efforts. This experimentation is performed in a remote area to successfully
execute surgical operations. Here, drones are also used to capture the live envir-
onment and broadcast it to the remote control system. Authors claim successful
execution of the medical operation. However, it has been observed that delay in
signal transmission has affected surgical activities. These activities and corre-
sponding results are important to observe and analyze. Zhang and Lu [91] have
surveyed recent advancements in the field of bionics, flexible actuation, sensing,
and intelligent control systems. This study has prepared a detailed limitation ana-
lysis of various relevant studies performed flexible, secure, and application-specific
research directions for the usage of drones in real-time scenarios. The clinical
medical applications are very useful in ensuring the potential capabilities of med-
ical robots. In these clinical applications, medical robots are found to be flexible for
usage especially power, intelligent control, and stiffness in medical operations and
services. These three factors are also considered as important influence factors in
soft medical robot history. Siegfarth et al. [92] discussed the usage of 3D-printed
hydraulic actuators for medical operations with medical robots. Such sensors and
actuators in the medical sector help in collecting real-time patient data and helpful
for immediate observations and treatment as well.

10.10.2 Drone and IoT network for health-care operations

This section discusses the usage of drone devices for health-care operations. There
are many approaches proposed in recent times for providing medical services
[34,87,88]. Internet of Drones (IoDs) in the case of multi-drone movement or IoD
and robot devices is necessary to successfully perform medical operations
[34,87,88]. Some of the recent development in this area is explained as follows.
214 The Internet of medical things

Kumar et al. [34] proposed a drone-based system for thermal imaging, saniti-
zation, monitoring, surveillance, and medication. In the proposed system, data can
be collected and processing at local and remote places before to share with the
hospital. Here, a drone-based system is designed and used for health-care opera-
tions. To avoid collision among multi-drone flying strategies, single and multilayer
drone flying approaches are proposed. Here, an analysis of the proposed approach
is made using simulators. Results show that the proposed approach is efficient in
drone flying and handling health-care operations. Lum et al. [87] proposed a drone
and robot integrated system for surgical operation. Here, a drone device is used to
move to a remote area where it is difficult to provide health-care service. There, a
robot is an instruction to operate and perform medical surgery. Here, a drone device
is used to signal and arrange the medical facilities necessary for medical operations.
Câmara et al. [88] have discussed the importance of drones in disaster scenarios. In
such scenarios, drone devices are very useful in remote monitoring and surveil-
lance. Drone devices can collect the live on-ground status that can help in arranging
necessary services especially medical and first aid kits. In today’s scenario, drone
devices are available for touchless screens and secure reporting. These services can
make pandemic handling much easier and simpler. The usage of such devices can
reduce the cost of monitoring and extending the health-care service for needful
compared to costs observed during COVID-19 times. Kim et al. [89] addressed the
issue of providing medication and test kits to chronic disease patients. This
experimentation has used drones for this service and it has been found that drones
are very useful in this case. Here, drones are used for two major purposes: (i)
identifying the optimal number of drones required for handling medical services
and (ii) how to reduce the cost of drone-based medical operations where medication
is provided to the patient and test kits (that include blood and other samples) are
collected. In this approach, optimization approach is used for cost–benefit analysis.
Results show that drone-aided health-care services are very powerful for health-
care services and easy to apply. Graboyes and Skorup [90] explored medical
drones. Here, technical and policy challenges are discussed in detail. Flying drones
and collecting data can breach privacy and security. Thus, flying drones in many
countries are not allowed. Besides, the lack of drone-related policies does not
provide easy and effective solutions for integrating this technology with the real-
time application such as healthcare. Here, the solution is proposed to have an initial
step to make it feasible on-ground. In conclusion, implementing the proposed
approach in the real-time application needs to solve large challenges that are dif-
ficult to achieve in the present scenario. To address this challenge, infrastructure
level changes and transparent policies are required.

10.10.3 Robot, drone, and IoT network integrations for

health-care applications
In [87–90], it has been observed that drone usages can be observed from inde-
pendent flying or observational planning. In [93–101], drones can be used in an
interconnected environment using IoT concepts. The use of IoT or constituting
 Patient-centric smart health-care systems 215

IoDs help in preparing a collaborative support environment with partial or full

automation. IoD can be controlled either through fixed cellular infrastructure or pri-
vately governed policies. In private IoD, Internet governance principles are required
to successfully implement it. In Internet governance principles, application-specific
nomenclature, addressing, interconnection, access control, and security primitives are
required to be clarified. For example, there are a large number of small- to large-size
medical equipment available for surgical operations [102]. Identification of equip-
ment, their operations (to maximum possibilities), patient monitoring, live image
capturing, contactless medication, and medical services can be extended with drones
and their collaboration. IoD is a concept that can ensure maximum security as well.
Presently, drone noise, large size, battery constraints, and lesser security are major
constraints to apply in real scenarios. However, outdoor drone services to rural and
remote areas can fulfill the need for lack of infrastructure and manpower to handle
health-care services. The major challenges in implementing drone-based strate-
gies in medical or health-care operations include the following: (i) air-space
awareness is required. For an electromechanical device, it is difficult to identify
the space where it has to execute its operations. Thus, a preplanned strategy is
required and its mapping in the drone devices is equally important. (ii) Non-line-
of-sight control is another major challenge. In this challenge, controlling the
drones across the boundaries is considered at large. In the case of IoD, the same
challenge exists as well. In IoD, drones in line-of-sight of each other can work
collaboratively and are controllable as well. It is difficult to successfully move,
control, and operate the drones in non-line-of-sight of control. (iii) Location of
drone movement is equally important. In a large number of countries, flying
drones in open space are not allowed because of security and privacy issues. Thus,
it is always better to preplan and take the government’s approval by ensuring that
flying will be safe and secure from every aspect. (iv) Finally, a collision-free
environment is required for drones to fly and operate. Collisions can be drone to
drone, and drone to object. Drone-to-drone collision avoidance comes in colla-
borative flying strategy in IoD. On the other hand, drone-to-object collision
avoidance is part of awareness of the area and preplanned flying strategy. In real
time, collisions can be avoided with LiDAR.

10.10.4 Recent developments and future directions

Kumar et al. [103] proposed a drone-based approach for handling the COVID-19
pandemic. This approach has proposed a single and multiplayer drone movement
strategy. In this preplanned strategy, multiple drones can fly without collision. To
avoid collisions, either the LiDAR approach can be used or a preplanned drone
movement strategy can be followed. With this approach, it is possible to provide
health-care services on a large scale as well. It has been observed that COVID-19
patients are increasing day by day. To handle a large number of patients with
minimum contact is possible through drones and robotics. Thus, a collision-free
drone movement environment is useful for handling COVID-19 and pandemic
situations. In the future, it is required to have drone-based data collection and
216 The Internet of medical things

real-time ML experimentation. ML experimentation would be the first step to

automate the drones for medical operations. Thereafter, it can be extended to have a
surgical drone or doctor drone. These drones can have collaborative efforts to
successfully perform multi-organ and multispecialty surgical operations. For
example, Li et al. [104] proposed multi-vehicle detection in a congested environ-
ment. With this approach, multi drone-based multi-objective detection will be
useful for health-care operations as well. McRae et al. [105] prepared the case
study to apply drones for complex high altitude search and rescue operations. Here,
the people and object detection approach can be applied to collect the data in real
time. This data can later be analyzed for process improvement and removing the
lags in observations. In pandemic situations, drones can be used at a large scale for
delivering blood, medicines, samples, and organs. Applying contactless organ
delivery would be a major challenge for drones but if it is found to be experi-
mentally successful, it would be the fastest and best way of saving lives. In the
present scenario, it has been observed that the lack of medical services in rural and
remote areas was a major hurdle to handle COVID-19 pandemic situations. With
drones, this can be made possible easily. In the present scenario, the drone market
is increasing day by day. A huge investment is expected to be prepared with multi-
drone multi-application planning. With this approach, drones can be effectively
used on a large scale.

10.11 Conclusion and future scope

This work has surveyed and explored the usage of information technology and its
recent developments in the health-care sector. Various technological advancements
made the usage of technology important for various applications, including health-
care and medical services. This work has explored the recent proposal in AI, ML,
drone, robotics, cryptography, and IoT to the health-care sector. This survey and
various comparative analyses drawn in this work are useful to analyze the factor
necessary while taking the advantage of technology in the health-care sector. Some
technologies (such as drones) require infrastructure level changes to accommodate
a collision-free and safe environment without which it would be difficult to adapt in
real-time application.

References

[1] Singh, R. P., Javaid, M., Haleem, A. and Suman, R., 2020. Internet of things
(IoT) applications to fight against COVID-19 pandemic. Diabetes &
Metabolic Syndrome: Clinical Research & Reviews, 14(4), pp. 521–524.
[2] Kamal, M., Aljohani, A. and Alanazi, E., 2020. IoT meets COVID-19:
Status, challenges, and opportunities. arXiv preprint arXiv: 2007. 12268.
[3] Amit Kumar M., 2019. IoT Healthcare, GitHub, https://github.com/mis-
hracodes/IOT-Healthcare.
 Patient-centric smart health-care systems 217

[4] Sharath S., HealthCareSystem IoT, GitHub, https://github.com/sharath29/

Health-Care-System-IOT.
[5] Drupad K., 2018. HealthCare IoT, GitHub, https://github.com/dkhublani/
Healthcare-IOT.
[6] Kien N., 2020. IoT HealthCare GitHub, https://github.com/Nyquixt/iot-
healthcare.
[7] Ritvik, 2017. IoT HealthCare GitHub, https://github.com/ritvik91/IoT-
healthcare.
[8] Mumtaz, S., Bo, A., Al-Dulaimi, A. and Tsang, K. F., 2018. Guest editorial
5G and beyond mobile technologies and applications for industrial IoT
(IIoT). IEEE Transactions on Industrial Informatics, 14(6), pp. 2588–2591.
[9] Rathee, G., Sharma, A., Kumar, R. and Iqbal, R., 2019. A secure commu-
nicating things network framework for industrial IoT using blockchain
technology. Ad Hoc Networks, 94, p. 101933.
[10] Simić, M., Sladić, G. and Milosavljević, B., 2017, June. A case study IoT
and blockchain powered healthcare. In Proc. ICET.
[11] McGhin, T., Choo, K. K. R., Liu, C. Z. and He, D., 2019. Blockchain in
healthcare applications: Research challenges and opportunities. Journal of
Network and Computer Applications, 135, pp. 62–75.
[12] Panarello, A., Tapas, N., Merlino, G., Longo, F. and Puliafito, A., 2018.
Blockchain and IoT integration: A systematic survey. Sensors, 18(8),
p. 2575.
[13] Reyna, A., Martı́n, C., Chen, J., Soler, E. and Dı́az, M., 2018. On blockchain
and its integration with IoT. Challenges and opportunities. Future
Generation Computer Systems, 88, pp. 173–190.
[14] Nichol, P. B., Blockchain applications for healthcare, March 2016 [online].
Available: http://www.cio.com/article/3042603/innovation/blockchain-appli-
cations-for-healthcare.html.
[15] Satamraju, K. P and Malarkodi, B., 2020. Proof of concept of scalable
integration of internet of things and blockchain in healthcare. Sensors, 20(5),
p. 1389.
[16] Singh, K., Kaushik, K., Ahatsham and Shahare, V., 2020. “Role and impact
of wearables in IoT healthcare,” in Advances in Intelligent Systems and
Computing, 1090, pp. 735–742. Singapore: Springer. doi: 10.1007/978-981-
15-1480-7_67.
[17] Kaushik, K. and Singh, K., 2020. “Security and trust in IoT communications:
Role and impact,” in Advances in Intelligent Systems and Computing,
vol. 989, pp. 791–798. Singapore: Springer. doi: 10.1007/978-981-13-8618-
3_81.
[18] Brantly, A. F. and Brantly, N. D., 2020. Patient-centric cybersecurity.
Journal of Cyber Policy, 5(3), pp. 372–391, doi: 10.1080/23738871.
2020.1856902.
[19] Lu, Y. and Da Xu, L., 2019. Internet of things (IoT) cybersecurity research:
A review of current research topics. IEEE Internet of Things Journal, 6(2),
pp. 2103–2115, doi: 10.1109/JIOT.2018.2869847.
218 The Internet of medical things

[20] Yaqoob, I., Ahmed, E., Muhammad Habib ur Rehman, et al., 2017. The rise
of ransomware and emerging security challenges in the Internet of Things.
Computer Networks, 129, pp. 444–458, doi: 10.1016/j.comnet.2017.09.003.
[21] Nasiri, S., Sadoughi, F., Tadayon, M. H. and Dehnad, A., 2019. Security
requirements of internet of things-based healthcare system: A survey study.
Acta Informatica Medica, 27(4), pp. 253–258, doi: 10.5455/aim.2019.
27.253-258.
[22] Usmonov, B., Evsutin, O., Iskhakov, A., Shelupanov, A., Iskhakova, A. and
Meshcheryakov, R., 2017. The cybersecurity in development of IoT
embedded technologies. In 2017 International Conference on Information
Science and Communications Technologies, ICISCT 2017, vol. 2017 (pp.
1–4), doi: 10.1109/ICISCT.2017.8188589.
[23] Mohan, A., 2014. Cyber security for personal medical devices internet of
things. In Proceedings – IEEE International Conference on Distributed
Computing in Sensor Systems, DCOSS 2014 (pp. 372–374), doi: 10.1109/
DCOSS.2014.49.
[24] Conti, M., Dehghantanha, A., Franke, K. and Watson, S., 2018. Internet of
Things security and forensics: Challenges and opportunities. Future Generation
Computer Systems, 78, pp. 544–546, doi: 10.1016/j.future.2017.07.060.
[25] Abie, H., 2019. Cognitive cybersecurity for CPS-IoT enabled healthcare
ecosystems. In International Symposium on Medical Information and
Communication Technology, ISMICT, vol. 2019, doi: 10.1109/
ISMICT.2019.8743670.
[26] Coventry, L. and Branley, D., 2018. Cybersecurity in healthcare: A narrative
review of trends, threats and ways forward. Maturitas, 113, pp. 48–52, doi:
10.1016/j.maturitas.2018.04.008.
[27] Pan, J. and Yang, Z., 2018. Cybersecurity challenges and opportunities in the
new ‘edge computing þ IoT’ world. In SDN-NFVSec 2018 – Proc.
2018 ACM Int. Work. Secur. Softw. Defin. Networks Netw. Funct.
Virtualization, Co-located with CODASPY 2018, vol. 2018 (pp. 29–32), doi:
10.1145/3180465.3180470.
[28] Marques, G., Pitarma, R., Garcia, N. M. and Pombo, N., 2019. Internet of
things architectures, technologies, applications, challenges, and future
directions for enhanced living environments and healthcare systems: A
review. Electronics, 8(10), p. 1081, doi: 10.3390/electronics8101081.
[29] Ahad, A., Tahir, M. and Yau, K. L. A., 2019. 5G-based smart healthcare
network: Architecture, taxonomy, challenges and future research directions.
IEEE Access, 7, pp. 100747–100762, doi: 10.1109/ACCESS.2019.2930628.
[30] Farahani, B., Firouzi, F., Chang, V., Badaroglu, M., Constant, N. and
Mankodiya, K., 2018. Towards fog-driven IoT eHealth: Promises and chal-
lenges of IoT in medicine and healthcare. Future Generation Computer
Systems, 78, pp. 659–676, doi: 10.1016/j.future.2017.04.036.
[31] Featherstone, R. M., Boldt, R. G., Torabi, N. and Konrad, S. L., 2012.
Provision of pandemic disease information by health sciences librarians: A
 Patient-centric smart health-care systems 219

multisite comparative case series. Journal of the Medical Library

Association: JMLA, 100(2), p. 104.
[32] Desai, B. C., 2020. Pandemic and big tech. In Proceedings of the 24th
Symposium on International Database Engineering & Applications
(pp. 1–10).
[33] Gupta, D., Bhatt, S., Gupta, M. and Tosun, A. S., 2021. Future smart con-
nected communities to fight Covid-19 outbreak. Internet of Things, 13,
p. 100342.
[34] Kumar, A., Sharma, K., Singh, H., Naugriya, S. G., Gill, S. S. and Buyya, R.,
2021. A drone-based networked system and methods for combating cor-
onavirus disease (COVID-19) pandemic. Future Generation Computer
Systems, 115, pp. 1–19.
[35] El-Shafai, W. and El-Samie, F. A. (12 June 2020), Extensive COVID-19 X-
Ray and CT Chest Images Dataset, Mendeley Data. https://data.mendeley.
com/datasets/8h65ywd2jr/3
[36] Khoong, W. H., (March 2020) COVID-19 Xray Dataset (Train & Test Sets),
Kaggle, https://www.kaggle.com/khoongweihao/covid19-xray-dataset-train-
test-sets/version/1
[37] Cohen, J. P., Morrison, P., Dao, L., Roth, K., Duong, T. Q. and Ghassemi, M.
COVID-19 Image Data Collection: Prospective Predictions Are the Future.
Retrieved from https://github.com/ieee8023/covid-chestxray-dataset, 2020,
arXiv:2006.11988.
[38] Alqudah, A. M. and Qazan, S. (26March 2020), Mendeley Data https://data.
mendeley.com/datasets/2fxz4px6d8/4
[39] Zhao, J., Zhang, Y., He, X. and Xie, P., 2020, COVID-CT-Dataset: a CT
scan dataset about COVID-19, Github, https://github.com/UCSD-AI4H/
COVID-CT
[40] Mangal, A., Kalia, S., Rajgopal, H., Rangarajan, K., Namboodiri, V.,
Banerjee, S. and Arora, C., 2020, CovidAID: COVID-19 Detection Using
ChestX-Ray, Github, https://github.com/arpanmangal/CovidAID
[41] Wang, L., Lin, Z. Q. and Wong, A., 2020, COVID-Net: a tailored deep
convolutional neural network design for detection of COVID-19 cases from
chest X-ray images, https://doi.org/10.1038/s41598-020-76550-z
[42] General Blockchain, 2021, covid-19-chest-xray-segmentations-dataset,
https://github.com/GeneralBlockchain/
[43] Sajid, N., 2020, COVID-19 Patients Lungs X Ray Images 10000, Kaggle,
https://www.kaggle.com/nabeelsajid917/covid-19-x-ray-10000-images
[44] Rahman, T., 2020, COVID-19 Radiography Database, Kaggle, https://www.
kaggle.com/tawsifurrahman/covid19-radiography-database
[45] Khoong, W. H., (March 2020) COVID-19 Xray Dataset (Train & Test Sets),
Kaggle, https://www.kaggle.com/khoongweihao/covid19-xray-dataset-train-
test-sets
[46] Jenssen, H. B., 2020, COVID-19 CT segmentation dataset, https://medium.
com/@hbjenssen/covid-19-radiology-data-collection-and-preparation-for-
artificial-intelligence-4ecece97bb5b
220 The Internet of medical things

[47] Rahman, T., 2020, COVID 19 Detection with X-Ray & COVID19 - Pytorch,
Kaggle, https://www.kaggle.com/portgasray/covid-19-detection-with-x-ray-
covid19-pytorch
[48] Società Italiana di Radiologia,2020, covid-19 Database, https://sirm.org/
category/senza-categoria/covid-19/
[49] Dr Daniel J. Bel , 30 Oct 2021, COVID-19 radiopaedia, https://radiopaedia.
org/articles/covid-19-4?lang=us
[50] Ma, J.; 2020, COVID-19 CT Lung and Infection Segmentation Dataset,
zendo, https://zenodo.org/record/3757476#.YX2FOlVBzIW
[51] Ma, J. et al., 2021, MP-COVID-19-SegBenchmark, gitee, https://gitee.com/
junma11/COVID-19-CT-Seg-Benchmark
[52] Born, J. et al., 2021, Accelerating Detection of Lung Pathologies with
Explainable Ultrasound Image Analysis ,GitHub, https://github.com/jannis-
born/covid19_pocus_ultrasound
[53] Stanislav E. S., 2020, Artificial intelligence in radiology, MosMedData,
https://mosmed.ai/en/
[54] Eduardo, P., 2020, SARS-COV-2 Ct-Scan Dataset, Kaggle, SARS-COV-2
Ct-Scan Dataset
[55] BIMCV-COVID19,2020, BIMCV-COVID19, Datasets related to
COVID19’s pathology course, BIMCV, https://bimcv.cipf.es/bimcv-pro-
jects/bimcv-covid19/
[56] Yang H. and Sohn E. Expanding Our Understanding of COVID-19 from
Biomedical Literature Using Word Embedding. Int J Environ Res Public
Health. 2021; 18(6): 3005. doi: 10.3390/ijerph18063005. PMID: 33804131;
PMCID: PMC7998313
[57] Banerjee, I., Sinha, P., Purkayastha, S., et al., 2020. Was there COVID-19
back in 2012? Challenge for AI in Diagnosis with Similar Indications. arXiv
preprint arXiv: 2006. 13262.
[58] Ahsan, M. M., Gupta, K. D., Islam, M. M., Sen, S., Rahman, M. and
Hossain, M. S., 2020. Study of different deep learning approach with
explainable AI for screening patients with COVID-19 symptoms: Using CT
scan and chest x-ray image dataset. arXiv preprint arXiv: 2007. 12525.
[59] Sujath, R., Chatterjee, J. M. and Hassanien, A. E., 2020. A machine learning
forecasting model for COVID-19 pandemic in India. Stochastic
Environmental Research and Risk Assessment, 34, pp. 959–972.
[60] Ambigai, S. D., Manivannan, K. and Shanthi, D., 2018. An efficient virtual
machine migration for smart healthcare using particle swarm optimization
algorithm. International Journal of Pure and Applied Mathematics, 118(20),
pp. 3715–3722.
[61] Demirkan, H., 2013. A smart healthcare systems framework. IT
Professional, 15(5), pp. 38–45.
[62] Abd EL-Latif, A. A., Abd-El-Atty, B., Abou-Nassar, E. M. and Venegas-
Andraca, S. E., 2020. Controlled alternate quantum walks based privacy
preserving healthcare images in internet of things. Optics & Laser
Technology, 124, p. 105942.
 Patient-centric smart health-care systems 221

[63] Fernández-Caramés, T. M., 2019. From pre-quantum to post-quantum IoT

security: A survey on quantum-resistant cryptosystems for the Internet of
Things. IEEE Internet of Things Journal, 7(7), pp. 6457–6480.
[64] Flöther, F., Murphy, J., Murtha, J. and D. Sow, Exploring quantum com-
puting use cases for healthcare. Accelerate diagnoses, personalize medicine,
and optimize pricing, IBM Report, URL: https://www.ibm.com/downloads/
cas/8QDGKDZJ [LastAccessed on: 31October,2021].
[65] Kumar, A., Gopal, K. and Aggarwal, A., 2013. Lightweight trust propaga-
tion scheme for resource constraint mobile ad-hoc networks (MANETs). In
2013 Sixth International Conference on Contemporary Computing (IC3)
(pp. 421–426). IEEE.
[66] Kumar, A., 2013. Performance & probability analysis of Lightweight
Identification Protocol. In 2013 International Conference On Signal
Processing And Communication (ICSC) (pp. 76–81). IEEE.
[67] Kumar, B., Prasad, S. B., Pal, P. R. and Pathak, P., 2021. Quantum security
for IoT to secure healthcare applications and their data. In Limitations and
Future Applications of Quantum Cryptography (pp. 148–168). Hershey, PA:
IGI Global.
[68] Fernández-Caramés, T. M. and Fraga-Lamas, P., 2020. Towards post-
quantum blockchain: A review on blockchain cryptography resistant to
quantum computing attacks. IEEE Access, 8, pp. 21091–21116. Piscataway,
NJ.
[69] Chaudhary, R., Jindal, A., Aujla, G. S., Kumar, N., Das, A. K. and Saxena,
N., 2018. LSCSH: Lattice-based secure cryptosystem for smart healthcare
in smart cities environment. IEEE Communications Magazine, 56(4),
pp. 24–32. USA.
[70] Li, Z., Wang, J. and Zhang, W., 2019. Revisiting post-quantum hash proof
systems over lattices for Internet of Thing authentications. Journal of
Ambient Intelligence and Humanized Computing, pp. 1–11. Germany.
[71] Alhadhrami, Z., Alghfeli, S., Alghfeli, M., Abedlla, J. A. and Shuaib, K.,
2017, November. Introducing blockchains for healthcare. In 2017
International Conference on Electrical and Computing Technologies and
Applications (ICECTA) (pp. 1–4). USA: IEEE.
[72] Yi, H., Li, J., Lin, Q., et al., 2019. A rainbow-based authentical scheme for
securing smart connected health systems. Journal of Medical Systems, 43(8),
pp. 1–10.
[73] Sepulveda, J., Zankl, A. and Mischke, O., 2017, September. Cache attacks
and countermeasures for NTRUEncrypt on MPSoCs: Post-quantum resis-
tance for the IoT. In 2017 30th IEEE International System-on-Chip
Conference (SOCC) (pp. 120–125). Munich: IEEE.
[74] Zhao, W. and Sampalli, S., 2020. Sensing and Signal Processing in Smart
Healthcare. Switzerland.
[75] Imran, M., Abideen, Z. U. and Pagliarini, S., 2020. An experimental study of
building blocks of lattice-based NIST post-quantum cryptographic algo-
rithms. Electronics, 9(11), p. 1953. Switzerland.
222 The Internet of medical things

[76] Li, C., Tian, Y., Chen, X. and Li, J., 2021. An efficient anti-quantum lattice-
based blind signature for blockchain-enabled systems. Information Sciences,
546, pp. 253–264.
[77] Akleylek, S., Goi, B. M., Yap, W. S., Wong, D. C. K. and Lee, W. K., 2018.
Fast NTRU encryption in GPU for secure IoP communication in post-
quantum era. In 2018 IEEE SmartWorld, Ubiquitous Intelligence &
Computing, Advanced & Trusted Computing, Scalable Computing &
Communications, Cloud & Big Data Computing, Internet of People and
Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/
SCI) (pp. 1923–1928). USA: IEEE.
[78] Bellini, E., Caullery, F., Hasikos, A., Manzano, M. and Mateu, V., 2018,
May. You shall not pass!(once again) an IoT application of post-quantum
stateful signature schemes. In Proceedings of the 5th ACM on ASIA Public-
Key Cryptography Workshop (pp. 1924). USA.
[79] Chen, J., Ling, J., Ning, J., et al., 2020. Post-quantum proxy signature
scheme based on the multivariate public key cryptographic signature.
International Journal of Distributed Sensor Networks, 16(4), p. 15501477
20914775.
[80] Gill, S. S., Kumar, A., Singh, H., et al., 2020. Quantum computing: A tax-
onomy, systematic review and future directions. arXiv preprint arXiv: 2010.
15559. UK: Software Practice and Experience, Wiley.
[81] Kumar, A., Elsersy, M., Darwsih, A. and Hassanien, A. E., 2021. “Drones
combat COVID-19 epidemic: Innovating and monitoring approach,” in
Digital Transformation and Emerging Technologies for Fighting COVID-19
Pandemic: Innovative Approaches, p. 175. Cham: Springer.
[82] Kumar, A. and Sharma, K., 2021. “Digital transformation and emerging
technologies for COVID-19 pandemic: Social, global, and industry per-
spectives,” in Artificial Intelligence and Machine Learning for COVID-19,
p. 73. Cham: Springer.
[83] Kumar, A., Sharma, K., Singh, H., Srikanth, P., Krishnamurthi, R. and
Nayyar, A., 2021. “Drone-based social distancing, sanitization, inspection,
monitoring, and control room for COVID-19,” in Artificial Intelligence and
Machine Learning for COVID-19, p. 153. Cham: Springer.
[84] Krishnamurthi, R., Gopinathan, D. and Kumar, A., 2021. “Wearable devices
and COVID-19: State of the art, framework, and challenges,” in Emerging
Technologies for Battling COVID-19: Applications and Innovations, p. 157.
Cham: Springer.
[85] Kumar, A. and Jain, S., 2021. “Drone-based monitoring and redirecting
system,” in Development and Future of Internet of Drones (IoD): Insights,
Trends and Road Ahead, p. 163. Cham: Springer.
[86] Sharma, K., Singh, H., Sharma, D. K., Kumar, A., Nayyar, A. and
Krishnamurthi, R., 2021. “Dynamic models and control techniques for drone
delivery of medications and other healthcare items in COVID-19,” in
Emerging Technologies for Battling Covid-19: Applications and
Innovations, p. 1. Cham: Springer.
 Patient-centric smart health-care systems 223

[87] Lum, M. J., Rosen, J., King, H. H., et al., 2007. Telesurgery via unmanned
aerial vehicle (UAV) with a field deployable surgical robot. In MMVR
(pp. 313–315). Netherlands.
[88] Câmara, D., 2014, November. Cavalry to the rescue: Drones fleet to help
rescuers operations over disasters scenarios. In 2014 IEEE Conference on
Antenna Measurements & Applications (CAMA) (pp. 1–4). Antibes Juan-les-
Pins, France: IEEE.
[89] Kim, S. J., Lim, G. J., Cho, J. and Côté, M. J., 2017. Drone-aided healthcare
services for patients with chronic diseases in rural areas. Journal of
Intelligent & Robotic Systems, 88(1), pp. 163–180.
[90] Graboyes, R. F. and Skorup, B., 2020. Medical Drones in the United States
and a Survey of Technical and Policy Challenges. USA: Mercatus Center
Policy Brief.
[91] Zhang, Y. and Lu, M., 2020. A review of recent advancements in soft and
flexible robots for medical applications. The International Journal of
Medical Robotics and Computer Assisted Surgery, 16(3), p. e2096.
[92] Siegfarth, M., Pusch, T. P., Pfeil, A., Renaud, P. and Stallkamp, J., 2020.
Multi-material 3D printed hydraulic actuator for medical robots. Rapid
Prototyping Journal.
[93] Gharibi, M., Boutaba, R. and Waslander, S. L., 2016. Internet of drones.
IEEE Access, 4, pp. 1148–1162.
[94] Lin, C., He, D., Kumar, N., Choo, K. K. R., Vinel, A. and Huang, X., 2018.
Security and privacy for the internet of drones: Challenges and solutions.
IEEE Communications Magazine, 56(1), pp. 64–69.
[95] Lv, Z., 2019. The security of Internet of drones. Computer Communications,
148, pp. 208–214.
[96] Kumar, A., Gopal, K. and Aggarwal, A., 2014. Design and analysis of
lightweight trust mechanism for accessing data in MANETs. KSII
Transactions on Internet & Information Systems, 8(3).
[97] Kumar, A., Gopal, K. and Aggarwal, A., 2013. Outlier detection and treat-
ment for lightweight mobile ad hoc networks. In International Conference
on Heterogeneous Networking for Quality, Reliability, Security and
Robustness (pp. 750–763). Springer, Berlin, Heidelberg.
[98] Kumar, A., Krishnamurthi, R., Nayyar, A., Sharma, K., Grover, V. and
Hossain, E., 2020. A novel smart healthcare design, simulation, and imple-
mentation using healthcare 4.0 processes. IEEE Access, 8, pp. 118433–118471.
[99] Kumar, A., Srikanth, P., Nayyar, A., Sharma, G., Krishnamurthi, R. and
Alazab, M., 2020. A novel simulated-annealing based electric bus system
design, simulation, and analysis for Dehradun Smart City. IEEE Access, 8,
pp. 89395–89424.
[100] Kumar, A., Aggarwal, A. and Gopal, K., 2018. A novel and efficient
reader-to-reader and tag-to-tag anti-collision protocol. IETE Journal of
Research, pp. 1–12.
[101] Kumar, A., Rajalakshmi, K., Jain, S., Nayyar, A. and Abouhawwash, M.,
2020. A novel heuristic simulation-optimization method for critical
224 The Internet of medical things

infrastructure in smart transportation systems. International Journal of

Communication Systems, 33(11), p. e4397.
[102] Kumar, A., Gopal, K. and Aggarwal, A., 2016. Simulation and cost analysis
of group authentication protocols. In 2016 Ninth International Conference
on Contemporary Computing (IC3) (pp. 1–7). IEEE.
[103] Kumar, A., Krishnamurthi, R., Nayyar, A., Luhach, A. K., Khan, M. S. and
Singh, A., 2021. A novel Software-Defined Drone Network (SDDN)-based
collision avoidance strategies for on-road traffic monitoring and manage-
ment. Vehicular Communications, 28, p. 100313.
[104] Li, W., Li, H., Wu, Q., Chen, X. and Ngan, K. N., 2019. Simultaneously
detecting and counting dense vehicles from drone images. IEEE
Transactions on Industrial Electronics, 66(12), pp. 9651–9662.
[105] McRae, J. N., Gay, C. J., Nielsen, B. M. and Hunt, A. P., 2019. Using an
unmanned aircraft system (drone) to conduct a complex high altitude
search and rescue operation: A case study. Wilderness & Environmental
Medicine, 30(3), pp. 287–290.
 Chapter 11
Application of intelligent techniques in
health-care sector
Niharika Singh1, Richa Choudhary1, Thipendra Pal Singh1
and Anshika Mahajan1

Artificial intelligence (AI) is more important in today’s technological world. It is a

cacophony of technology. AI has been instrumental in transforming several aspects
of healthcare and has proven to be more efficient than human caregivers. The
dynamic increase in the number of population of the world provides excess pres-
sure over the health-care system. Hence, AI provides new technologies for deli-
vering the benefits to human health and well-being. Thus, this chapter focuses on
the importance of AI, health-care system in India followed by the future scope and
the challenges.

11.1 Introduction

AI is the emulation of human intelligence processes by computers. It is a term used

to characterize computing technologies that are analogous to human intelligence
processes, including “learning and adaptation,” “sensory understanding and inter-
action,” “reasoning and planning,” “searching and optimization,” and “autonomy
and creativity” [1]. Depending upon the domains, AI is classified as weak or strong.
Weak AI is a type of AI that is developed and educated for a specific job, such as
voice-activated assistants. It can either respond to your inquiry or carry out a pre-
programmed order, but it cannot function without human contact. Strong AI is a
type of AI that has generalized human cognitive skills, which means it can solve
problems and discover answers without the need for human interaction. A self-
driving automobile is an example of powerful AI that utilizes a mix of computer
vision, image recognition, and deep learning to steer a vehicle while keeping in a
specified lane and avoiding unforeseen hazards such as pedestrians.
Healthcare, education, finance, law, and manufacturing are just a few of the
areas where AI is being used to benefit both businesses and customers. Automation,
machine learning (ML), machine vision, natural language processing, and robotics

1
School of Computer Science, University of Petroleum and Energy Studies, Dehradun, India
226 The Internet of medical things

are just a few of the technologies that use AI. The use of AI presents legal, ethical,
and security problems. Empower new technology to enhance human health and
well-being in areas such as “primary care, service delivery, medical data integra-
tion and analysis, and disease outbreaks and other medical crises.” Projects are
intended to offer 1 billion additional individuals with improved healthcare and
well-being AI applications [2].
One of the driving forces behind the rise of AI is the data economy [3]. It
relates to how much data has grown in recent years and how much more can be
added in the future. The proliferation of data has spawned a new economy, and
there is a continual struggle for data ownership among businesses seeking to
profit from it. The increase in data volume has given rise to big data, which aids
in the management of massive volumes of data. Data science aids in the analysis
of the data. As a result, data science is moving toward a new paradigm in which
robots may be taught to learn from data and provide a range of useful insights,
giving rise to AI. AI refers to artificial intelligence that mimics human and animal
intelligence. It incorporates intelligent agents, which are self-contained creatures
that sense their surroundings and take actions to improve their chances of
achieving a certain objective. AI is a method for computers to imitate human
intellect through reasoning. It is a program with the ability to perceive, reason,
and act. AI is redefining industries by providing greater personalization to users
and automating processes.
ML is at the heart of AI. The first part of ML involves using algorithms to
discover meaning in random and unordered data, and the second part involves using
learning algorithms to identify a link between that knowledge enhance the learning
process. As a result, the ultimate objective of ML is to enhance the computers’
performance on a given job, such as diseases diagnosis, prediction of drugs, and
automation of healthcare services.
The causes for the rapid development may be ascribed to: an ageing popu-
lation need more health care services; an increase in chronic illnesses necessi-
tating more long-term healthcare services; and many older individuals prefer to
live in residential care facilities or in their own homes. These are not insignificant
reasons for the need for additional nurses, but the robotic revolution powered by
AI will be able to address them [4]. To alleviate this deficit in the medical area,
nursing robots [5] are already in use medical treatments that are more effective
and safer.

11.2 Evolution of AI in health-care informatics

Patient safety has long been a major concern in public health across the world.
There is a good possibility that a patient may experience nonessential damage
while undergoing therapy. About half of them could have been avoided. It is one of
the major contributors to the worldwide illness burden. Diagnostic errors are
common throughout medical care, according to the Institute of Medicine research,
and there is no effective way to reduce them. AI systems can give differential
 Application of intelligent techniques in health-care sector 227

diagnosis and helpful testing suggestions as a useful tool for identifying various
diseases. Medical providers and patients can benefit more from the huge medical
data of physiology, behavior, laboratory, and medical imaging, which is combined
with the development of ML technology.
Today, hospitals generate abundant data that provides analytics in order to
assist patients in the future which is extremely challenging. AI is a wide phrase that
refers to a large-scale endeavor to develop nonhuman intelligence. AI allows using
data from tens of thousands of patients to anticipate what will happen to a specific
patient and prevent it from occurring. To do so, researchers may utilize data from
tens of thousands of patients to predict who is at danger and intervene to prevent it
from happening. In healthcare, humans are dealing with an enormous quantity of
data, and the only way it can be evaluated is via the use of machines. The ultimate
objective of this type of technology is to reduce the cost, improve efficiency, and
ensure the safety of the service provided. The importance of early therapy in
enhancing outcomes cannot be overstated. Another area where it is beneficial is in
solving the global shortage of medical knowledge. AI’s contribution to healthcare
is that it can aid in the scaling up of some of these complicated and vital jobs. When
extra labor is necessary, which is common in the medical profession, AI models
work ceaselessly, which is a problem.
Another difficult deep learning topic with much more complicated data is
pathology [6]. Being able to visually distinguish between hundreds of diseases is
quite difficult. Additionally, field that has benefited greatly from deep learning
is genomics [7]. Medical data processing is demanding and hard to perform data
modeling. This may be accomplished by correcting bias in the training data as well
as bias in the model architecture and issue formulation. If none of this applies,
researchers may test and guarantee that equitable results and resource allocations
are achieved at the conclusion of AI model deployment. When it comes to public
health, epidemiological models are highly useful, but there are other aspects to
consider, such as climate change, flood forecasts for public health warnings, land
stability, and climate restrictions.

11.3 Healthcare in India

AI has brought a change in industries all around the world and one among those in
the health-care industry. AI is a superset of various concepts and technologies like
deep learning, ML, and natural language processing and integration of these tech-
nologies in digital healthcare can help doctors all around the world to achieve great
accuracies and better results. Moreover, discussions over digital healthcare have
increased due to the global pandemic COVID-19.
From predicting health problems in patients by processing and analyzing
their previous health data, to smart wearable fitness watches, AI plays a vital role
in the health-care sector. Governments of various countries and also major tech
giants are investing huge amounts into digital health. Earlier in the 1990s, Food
and Drug Administration (FDA)-approved algorithms were used to detect cancers
228 The Internet of medical things

in medical images but now new AI solutions and image detection techniques
have made these efforts much easier. The use of various ML techniques to train
the models from labeled data where inputs and outputs both are known and then
based upon the learning, the model predicts output feature for new test data.
The model keeps on learning and increases its efficiency to give better results.
This automation in the health industry can change the facet of the health industry
by many folds. To provide this automation, one needs to process the health-care
datasets for a well-defined research problem and the system can be trained
for automation.

11.4 Health-care dataset

Health-care data means any type of data related to the overall health and wellness
of an individual. Due to the huge advancement in computational techniques and
devices, a huge set of data is generated daily which can be analyzed to predict/
diagnose the current health issues. Table 11.1 shows the various forms of medical
data and its types along with how this data can be preprocessed to be able to feed to
the AI algorithm.
Health-care data is varied in nature and is difficult to process as it comes from
multiple sources and a variety of data types and forms. Implementing an AI algo-
rithm to predict or diagnose a particular health issue needs huge datasets which can
be converted into various labeled classes to detect the problem. Currently, available

Table 11.1 Different types of medical data

Medical data Data type Data preprocessing

1 X-rays Image Image is converted into pixel
for processing
2 Computed tomography (CT) Image Image is converted into pixel
for processing
3 Magnetic resonance imaging (MRI) Image/video Image is converted into pixel
for processing
4 Positron emission tomography (PET) Image Image is converted into pixel
for processing
5 Ultrasound Image/video Image is converted into pixel
for processing
6 Medical reports (blood/urine tests) Textual Normalization of data
7 ECG/EEG/PPG Signals Signal to image and pixels
8 Clinical trials (treatments) Textual Normalization of data
9 Health and fitness device Textual/signals/ Normalization of data for
numerical processing
10 Health insurance and claims Textual Normalization of data for
processing
11 Health surveys Textual Normalization of data for
processing
 Application of intelligent techniques in health-care sector 229

health-care datasets/databases are small and cannot fulfill the variety of cases that
exist across the huge population in India.

11.4.1 Electronic medical records

Electronic medical records (EMRs) include X-rays, ultrasounds, medicinal pre-
scriptions, lab records, etc., and these are being collected in mass volumes every-
day. Record of patients is an important factor, as it is used to learn about the health
of the patient and improve the type of treatment and medicines that the patient
requires. Various AI techniques are used in analyzing EMRs and provide adequate
information to doctors.
Large amounts of data stored in these EMRs are used to train various AI
algorithms. While training, these algorithms try to find some patterns in the data
and then generate certain rules, so that when next time new data of any patient is
provided, then the algorithm can evaluate the output using its experience from the
previous data. Data processing and its challenges are how to acquire, transform,
storage, and security of data. One of the biggest concerns while using health-care
data is acquiring the data itself. There are three major challenges while acquiring a
high-quality health-care dataset.

11.4.2 Reluctant to adopt EMR/digitalization of medical

procedures
In India, there are guidelines to collect EMR but no policy due to which many of them
are reluctant to start with EMR. Due to this, the practice of EMR is very limited.
Variety in data: No single standards or policies are in place for the medical
fraternity due to which available data is heterogeneous, complex, and difficult to
integrate.
Privacy and security concerns: Health is a personal thing so is the data
related to it. This belongs to the individuals. Harnessing this data has privacy and
security concerns, which is also a big challenge to address.
Indian Government is working in this direction and few initiatives are starting
to tackle these issues. One such initiative is open-source websites that collect and
maintain health-related data of Indian citizens by the health ministry of India. It is
called the Open Government Data platform of India. Such initiatives can provide a
baseline to improve the research in health-care field.
AI is amalgam of many other techniques and technologies, to effectively use
AI in health-care data processing, the following steps can be taken:

1. defining research question;

2. acquiring health-care dataset to be analyzed;
3. selecting AI algorithm (from plethora of Al algorithms) to be used for the
decided problem statement;
4. code to implement AI algorithm; and
5. analyzing, validating, and identifying future work.
230 The Internet of medical things

11.5 AI in healthcare
AI is a technology where machines analyze and process the data in way much
similar to human brain mechanism. AI is a big term that includes many other
techniques and technology to simulate the capabilities of human brain. This is a
promising technology that holds the key to the future of humanity [8,9]. It has
huge potential in all domains of life. Healthcare is one of the domains where AI
is most sought after technology due to the global pandemic—COVID-19 the
world is currently facing. Scientists and researchers are decoding AI to answer
the medical research questions that cannot be answered by any other
technologies.
AI can change the landscape of healthcare and is coloring the dream image of
revolutionizing healthcare and help address some of the challenges in the field.
Many countries are investing huge amount in health-care research using AI. India is
also investing in this field since 2012, which has also given rise to many health-care
startups in India [3]. India is a country that provides some of the best hospitals and
highly qualifies medical experts and staff; but still it lacks in proper health-care
infrastructure. There are many initiatives in collaboration with government, big
health-care companies, and hospitals to improve the health-care infrastructure in
India using AI techniques. Currently, there are many firms that are using AI tech-
niques to help in diagnosis and prediction of diseases. One such example is a
startup based in Bangalore which is using ML to diagnose cancer; few examples are
there who are trying to provide primary care and early detection of diseases based
on patients’ previous EMR. Seeing the huge potential of AI in health-care tech-
nology giants Google, Microsoft, IBM, and many others are collaborating with
govt. and health service providers to design software and tools to aid in health-care
services. AI in healthcare around the globe has many success stories and holds
promising results. Figure 11.1 shows the various medical fields where AI can
be used.
The applications mentioned earlier, some are already in use or in their early
phases of commercialization around the globe. But there are few misconceptions
about this high end technology—one of the most prevailing is—it will kill human
jobs or in terms of healthcare it will replace the medical staff or doctors. It is a
misconception, as there are still many areas of AI which have many challenges. It is
very far from replacing doctors or medical staff [4].
Currently, AI algorithms are lacking to fulfill the requirements of processing
any type of health-care data. It needs lot of efforts and big initiatives to build a self-
learning model that should be able to treat and manage diseases and can help in
clinical decision-making, it needs human intervention at many stages [5]. But
collaborative efforts of govt. big firms and medical fraternity can produce the
promising results in coming years.
The promising application of AI in health-care sector in current years is in
imagery detection, for example, cancer detection from the images and retinal
diseases detection where it provides great accuracy and very quick and precise
results.
 Application of intelligent techniques in health-care sector 231

Early diagnosis can take place

accessing the cumulative data of
a patient

Diagnosis Early diagnosis–early–

treatment–better results.
Improves efficacy of diagnosis;
avoid medical errors.

Artificial
intelligence

Huge dataset to analyse the Treatment stratergies

patterns for future according to the individuals.
predictions
Prediction Treatment Assistance for surgeries for
Huge capacity to store and greater precision.
process big data which
enhances the accuracy Virtual sssistance to remote
ofprediction. locations.
All healthcare devices
integration for better
understanding an individuals
health.

Automation of administration.
Improve patient experience.

Figure 11.1 Applications of AI in health-care domain

Another example is IBM Watson—it has healthcare-related question–answer-

ing computer system, which has health-care applications. IBM Watson helps
medical practitioners in making decisions by using natural language processing,
hypothesis creation, and evidence-based learning.

11.5.1 What could be achieved using AI in healthcare?

AI is making strides into the public health sector, and it has a significant impact on
all aspects of primary care. Primary care providers can better identify patients
who require extra attention and give tailored treatments for each one, owing to
AI-enabled computer tools [11,12].
1. Drug discovery: Discovery and the development of drugs takes several years
and costs a lot of money, but these efforts, time, and money can be reduced by
using various AI techniques. AI cannot complete the whole process of drug
discovery but can help in discovering new compounds and medicine formulas
that may form the desired drug. AI solutions can also help in determining more
and more applications of drugs that have been discovered and tested earlier [8].
New AI solutions are even being developed to determine new therapies of
medicines from databases related to medicines.
2. Dermatology: Deep learning that is a subset of AI mainly focuses on image
detection techniques and has contributed a lot in image processing.
Dermatology mainly depends upon images, and convolutional neural networks
232 The Internet of medical things

have achieved very high accuracies in determining skin cancer as well as other
skin diseases from various contextual, micro- and macro images [9].
3. Designing treatment plans and diagnosis: AI can help in designing pre-
scriptions and treatment plans for patients. AI can analyze data from other
patients and suggest strategies for treating other patients. From medical images
such as X-rays, MRIs, and ultrasounds, AI has higher capability to detect signs
of disease based upon various algorithms that suggest ways to diagnose them
as well. For example, the use of support vector machines (SVM) to detect heart
cancer based upon details of cell and use of neural networks to detect lung
cancer.
Following are ML techniques that are used [13]:
1. Supervised learning: In this kind of learning, model is provided with labeled
data. Inputs and outputs both are known, and the model tries to learn through
the given data. This data used to train the model is known as train data. After
learning the model tries to predict output values and if they are wrong, model
tries to improve the accuracy.
2. Un-supervised learning: In this kind of learning, input to the model is fed
with unlabeled data. The model is not provided with any relation between
inputs and outputs. This learning technique is very complex as model has to
learn itself. Unsupervised learning is mainly used for cluster analysis, anomaly
detection, etc.
The most immediate need for AI in healthcare is disease diagnosis. Early
diagnosis of common diseases such as breast cancer, diabetes, coronary artery
disease, and tumors can help patients control and minimize their chances of dying
from these illnesses. With the advancement in ML and AI, several classifiers
and clustering algorithms, including SVM, logistic regression/linear regression,
K-nearest neighbor, decision trees, Naı̈ve Bayes, random forest, and adaptive
boosting, are used for the prediction of the abovementioned diseases and gain
unprecedented insights into diagnostics [11].
As discussed earlier, health-care dataset comes in the form of image/video
and text/signals which is quite challenging to process and predict patterns. The AI
algorithms need to classify the available dataset into different classes. These
classes serve as training dataset to effectively recognize medical events or pat-
terns in the data. These algorithms need to be trained for any outliers as well for
better efficacy.

11.5.2 Challenges of using AI and possible solutions

In domains such as imaging and diagnosis, risk analysis, lifestyle management and
monitoring, health information management, and virtual health help, all kinds of AI
are being intensively explored for innovative health-care applications [10]. And the
expected proximate benefits of AI in these fields are numerous. Despite the
potential benefits of faster and more customized treatment and services, as well as
cheaper health-care costs, AI technologies also carry with them complex
 Application of intelligent techniques in health-care sector 233

organizational problems and long-term societal consequences. In the realm of

healthcare, the potential negative effects of AI on society, whether they are mag-
nifying human biases and injustice or the potential dangers of over-automation,
cannot be ignored since they directly impact decision-making about people’s
health. AI-based health-care technology, for example, might open up a plethora of
new scenarios in which generally held values and ethical standards are challenged.
The reliance on enormous volumes of health-related data is one of the most obvious
characteristics of AI in healthcare. Patient data acquired through wearable devices
or other types of surveillance to aid in the development of more customized and
individualized therapies might have a significant influence on consent procedures
such as notifying patients, asking for their preferences, and getting their agreement
[14]. Consent is critical when it comes to developing trustworthy connections
between patients/clients, physicians, and health institutions. To reflect the evolving
nature of health information and its consequences for patients/clients, consent
processes and documentation may need to be revised. To overcome this barrier,
both health institutions and technology firms must be open and honest about the AI.
A large set of challenges involves privacy and ethical data processing [15], as
well as the unclear future applications of health information beyond basic clinical
treatment. Diagnoses and prognoses may improve in accuracy and information
quality as a result of the insights gained from patient data. However, data misuse,
whether through cyber-hacking or by governing bodies, is a risk that can sway
patients’ opinions, hopes, and fears. To address developing privacy and data
handling problems, health organizations and practitioners may need to familiarize
themselves with new future rules for information security and data management as
well as provide employee training and create awareness about the issues. These
challenges must be solved if AI-driven technologies are to complement human
thinking and decision-making.
Integrating AI into health-care demands contact between AI and a broad set of
people, yet it can be expected to have some prejudice from AI users due to popular
culture’s notion of “scary AI.” The answer to this problem will be to raise public
knowledge of AI, dispel common misconceptions about the technology, and be able
to communicate concepts to governments in a simple and compelling manner so
that they understand the significance of adopting the technology.

11.5.3 Future scope

AI will play an important role in future health-care solution. It is a critical capacity
that underpins the development of precision medicine, which is widely recognized
as a much-needed breakthrough in healthcare. Most radiology and pathology pic-
tures are likely to be examined by a computer at some point in the future, thanks to
substantial advances in AI for imaging analysis [16,17]. Speech and text recogni-
tion are already being utilized for patient communication and clinical note
recording, and this trend is expected to grow. In many health-care disciplines, the
most challenging obstacle for AI is ensuring its acceptability in everyday clinical
practice, rather than whether the technology will be capable enough to be
234 The Internet of medical things

beneficial. These challenges may look daunting, but they might be an opportunity
for growth for organizations wanting to embrace exemplary responses to the clin-
ical and ethical problems raised by a tough but promising set of AI-based health-
care practices. Rapid advances in AI research, as well as government and private-
sector funding, make it extremely probable that AI will be widely used in health-
care delivery, with significant cost-cutting and service-quality-improvement
potential. In current scenario, huge amount of work needs to be done in the realm
of AI in health-care. Government needs to lay down guidelines in terms of privacy
and security, law and responsibility, psychological and ethical issues as far as AI in
healthcare is concerned. AI will become a strong tool for saving lives and
improving their quality of life if humans develop on all of these areas.

11.5.4 Current status of AI in healthcare

AI research in medicine is quickly expanding, and several attempts to create and
market AI-based medical devices have been launched. Furthermore, prominent
global information technology (IT) companies like Samsung, Google, Apple,
Microsoft, and Amazon, as well as a slew of competing startups, have demonstrated
significant research accomplishments in the application of AI in healthcare [18,19].
These initiatives by industry and medicine are assisting regulatory agencies in
licensing AI-based medical devices. The FDA in the United States initially allowed
the use of AI-based medical devices in 2017, while the Ministry of Food and Drug
Safety in Korea has approved the use of AI-based medical devices since 2018.

Table 11.2 Current applications of AI in healthcare [12]

Technology Description Application area

Machine learning It predicts the patterns in medical data to Disease diagnosis
give tailored made treatment, reduces
the ambiguity in patient treatment,
empowers decision-making by using
self-learning model to evaluate huge
data of medical images
Robotics Helps to perform surgeries with better Medical tools and
precision and accuracy devices
Natural language Converts text data, such as medical charts, Medical device,
processing into something that can be read and health IT
understood quickly
Big data analysis By analyzing large quantities of data Medicine, health IT
maintained by health-care organizations,
provides individualized recommenda-
tions to patients and medicines
Image processing Processes vast numbers of medical images Diagnostic medical
quickly and uses the data to determine image, health IT
disease kind, as well as negative and
positive test results
 Application of intelligent techniques in health-care sector 235

AI is a game changer in the medical field. It effectively analyzes data, medical

records, and systems, as well as improving digital automation to produce faster and
more trustworthy outcomes. AI has a significant capability in the medical business
to complete required activities with minimum human interaction. AI enables a
computer to comprehend human speech and writing in order to operate a business.
It gives doctors, surgeons, and physicians real-time advice on how to improve
results and perfect the skill [20,21]. AI provides step-by-step guidance and addi-
tional analysis to the surgeon in order to improve and get better results. For clinical
judgment, analysis, and training, AI looks to be the most successful tool. Moreover,
a number of research on the application of AI-based technologies in healthcare are
presently underway (Table 11.2).

References
[1] Secinaro, S., Calandra, D., Secinaro, A., et al. The role of artificial intelli-
gence in healthcare: a structured literature review. BMC Medical Informatics
and Decision Making 2021; 21: 125. https://doi.org/10.1186/s12911-021-
01488-9.
[2] Mead, L. Global Summit Focuses on The Role of Artificial Intelligence in
Advancing SDGs. SDG knowledge hub 2018. [accessed 10 August 2019].
Available at: http://sdg.iisd.org/news/global-summit-focuses-on-the-role-of-
artificial-intelligence-inadvancing-sdgs/.
[3] Vijai, C and Wisetsri, W. Rise of artificial intelligence in healthcare startups
in India, Advances in Management, 2021; 14(1): 48–52.
[4] Pepito, J. A. and Locsin, R. Can nurses remain relevant in a technologically
advanced future? International Journal of Nursing Sciences 2019; 6(1):
106–110 [accessed 20 November 2019]. Available at: https://www.science-
direct.com/science/article/pii/S2352013218301765.
[5] Väänänen, A., Haataja, K., Vehviläinen-Julkunen, K. and Toivanen, P. AI in
healthcare: a narrative review [version 1; peer review: 1 not approved].
F1000Research 2021; 10: 6. https://doi.org/10.12688/f1000research.26997.1.
[6] Yu, K.-H. , Beam, A. L. and Kohane, I. S. Artificial intelligence in health-
care. Nature Biomedical Engineering 2018; 2: 719–731. www.nature.com/
natbiomedeng.
[7] Kohli, P. S. and Arora, S. Application of machine learning in disease predic-
tion. 2018 4th International Conference on Computing Communication and
Automation (ICCCA), 2018. 978-1-5386-6947-1/18/$31.00 ’2018 IEEE.
[8] Claire Muñoz P. and Aneja, U. Artificial Intelligence for Healthcare:
Insights from India, ISBN: 978 1 78413 394 8, 2020. [Accessed 27 July
2021]. Available at: https://www.chathamhouse.org/2020/07/artificial-
intelligence-healthcare-insights-india.
236 The Internet of medical things

[9] Teneva, M. Debunking 10 Misconceptions about AI 2019. [Accessed 25 July

2021]. Available at: https://365datascience.com/trending/debunking-mis-
conceptions-ai/.
[10] Rong, G., Mendez, A., Assi, E. B., Zhao, B. and Sawan, M. Artificial
intelligence in healthcare: review and prediction case studies. Engineering
2020; 6(3): 291–301. ISSN 2095-8099, https://doi.org/10.1016/j.eng.2019.08.015.
(https://www.sciencedirect.com/science/article/pii/S2095809919301535).
[11] Manne, R. and Kantheti, S. C. Application of artificial intelligence in
healthcare: chances and challenges. Current Journal of Applied Science and
Technology 2021; 40(6): 78–89 67947 ISSN: 2457-102.
[12] Paul, D., Sanap, G., Shenoy, S., Kalyane, D., Kalia, K. and Tekade, R. K.
Artificial intelligence in drug discovery and development. Drug Discovery
Today 2021; 26(1): 80.
[13] Mantzaris, D. H., Anastassopoulos, G. C. and Lymberopoulos, D. K.
Medical disease prediction using artificial neural networks. 2008 8th IEEE
International Conference on BioInformatics and BioEngineering, 2008.
doi:10.1109/bibe.2008.4696782.
[14] Racine, E., Boehlen, W. and Sample, M. Healthcare uses of artificial intel-
ligence: Challenges and opportunities for growth. Healthcare Management
Forum 2019; 32: 272–275. doi:10.1177/0840470419843831.
[15] Schönberger, D. Artificial intelligence in healthcare: a critical analysis of the
legal and ethical implications. International Journal of Law and Information
Technology 2019; 27: 171–203. doi:10.1093/ijlit/eaz004.
[16] Jiang, F., Jiang, Y., Zhi, H., et al. Artificial intelligence in healthcare: past,
present and future. Stroke and Vascular Neurology 2017; 2: e000101.
doi:10.1136/svn-2017-000101.
[17] liashenko, O., Bikkulova, Z. and Dubgorn, A. Opportunities and challenges
of artificial intelligence in healthcare. E3S Web of Conferences 2019; 110:
02028. doi:10.1051/e3sconf/201911002028.
[18] Park, S. Y., Kuo, P.-Y., Barbarin, A., et al. Identifying challenges and
opportunities in human-AI collaboration in healthcare. Conference
Companion Publication of the 2019 on Computer Supported Cooperative
Work and Social Computing – CSCW ’19, 2019. doi:10.1145/3311957.
3359433.
[19] Reddy, S., Fox, J. and Purohit, M. P. Artificial intelligence-enabled health-
care delivery. Journal of the Royal Society of Medicine; 112(1) 1–7.
doi:10.1177/0141076818815510.
[20] Panesar, A. (2021). Machine Learning and AI for Healthcare. doi:10.1007/
978-1-4842-6537-6.
[21] Brinker, T. J., Hekler, A., Utikal, J. S., et al. Skin cancer classification using
convolutional neural networks: systematic review. Journal of Medical
Internet Research 2018; 20(10): e11936. Published 2018 Oct 17.
doi:10.2196/11936.
 Chapter 12
Managing clinical data using machine learning
techniques
V. Diviya Prabha1 and R. Rathipriya1

Analyzing clinical data is a great challenge in today’s digital data world. This
perspective imposes the need of machine learning (ML) algorithms to extract useful
patterns in clinical data. This chapter improves patient care by diagnosing disease
accurately. It also helps to study the importance of clinical data and managing into
PySpark environment. Various disease datasets are trained to ML techniques
(MLT) to identify the best model. It ensures the collection of clinical data from
different sources, integrating and extracting the useful patterns with less time
consumption. This approach improves the understanding of clinical data and
improves patient care.

12.1 Introduction
The promising approach of ML in health-care data is improving day by day. It is an
important technique that drives to the advancement of artificial intelligence.
Maintaining the history of patient’s data improves the health-care populations and
minimizes cost of healthcare. Nevertheless, the ability to manage huge datasets is
difficult for human knowledge and it is not reliable to convert or analyze medical
data. Converting the data into useful sights and attaching to real-time data is tedious.
Managing clinical data information and achieving a good decision is difficult
in health-care scenario. Extracting useful information to make a decision with
traditional man-made analysis is not sufficient. Medical data analysis is moved to
efficient analysis to promote a useful information and diagnose meaningful infor-
mation. The objective of this chapter is to understand that MLT concepts are
suitable for clinical data analysis. The prediction of disease might be different
based on features. This targets the improvement of individual patient care and
suggests the supporting features. For transforming clinical data [1] to useful data,
this approach is suitable. A novel ML algorithm [2] is developed for diagnosing
disease. The main objective of this chapter is to develop and classify MLT for

1
Department of Computer Science, Periyar University, Salem, India
238 The Internet of medical things

different diseases. The histories of patient records of various diseases such as dia-
betes, kidney and heart disease are diagnosed through MLT.
The objective of this work is to analyze clinical data and make all the models
run in pipeline process for datasets. Section 12.2 discusses the related work about
clinical data analysis in MLT. Section 12.3 comprises the results of clinical datasets
validation and results. Section 12.4 concludes the chapter.

12.2 Related work

Different ML algorithms are used for classification, which help to predict the diag-
nosis of different diseases. Different disease benchmark datasets are taken from the
UCI and trained using ML algorithm [3] to get better and more reliable prediction.
Running one algorithm and not comparing with any other algorithm does not show us
the best result. Finding a best classifier depends on comparing the algorithm with
more than one algorithm. It must be trained and tested with different datasets in a
health-care scenario. Ebola virus disease [1] prediction using ML is trained with
several predictors to qualify the best model. Also in predicting hospitalization, ML
plays an important role [4] to improve the clinical data as early prediction reduces the
patient risk. MLT is integrated into certain parts of health-care sectors to assist
health-care providers and patients [5]. It provides solution for challenging health-care
problems. However, complex use of methods [6] and opportunities are discussed. It
highlights to deal with clinical data toward technology. Addressing the usage [7] of
MLT in disease predictions suggests methods to improve patient care. With the rapid
progress of large amount of data [8–12], these techniques decrease the patient risk.
EHR (Electronic Health Record) [13] prediction model suggests MLT for clinical
decision for better patient outcome (Table 12.1). Table 12.1 describes the detail
description about the clinical dataset collected from UCI repository.

12.3 Clinical data analysis

The main objective of this chapter is to extract meaningful patterns from clinical
datasets collected from UCI repository. It helps us to take a better decision in
choosing the model and leverage to real-time data. The coding is developed in
PySpark. Figure 12.1 represents the flow of clinical records applied to MLT
approach using traditional techniques.

Table 12.1 Datasets details

Datasets Features Instances

Pima Diabetes [14] 9 769
Heart Disease [15] 14 304
Diabetes 130 US Hospital Dataset [16] 55 100 000
Kidney Disease [17] 26 401
 Managing clinical data using machine learning techniques 239

Clinical Records

Diabetes Train

Heart Machine
Data Feature learning
preprocessing selection techniques
Kidney
Test
Other
clinical
records

Figure 12.1 Clinical data analysis using MLT

ML pipeline
Process
LR
Train
Model
PySpark RF validation
Data frame and
source Test selection of
model

Other model

Figure 12.2 PySpark clinical data analysis

Figure 12.2 represents the flow of clinical data analysis in PySpark framework;
the data source is divided into training and testing. To develop a spark environment
[18] create a Spark Context to connect with clusters. Using the one hot encoding
method converts the categorical data into numerical data, String Indexer transforms
the string features into numerical form, for example, “M” represents that the male
is converted into 0 and similarly “F” represents the female as 1. Certain techniques
are used for preprocessing. The data is transformed to the pipeline to run the
algorithm simultaneously to achieve better results.

12.3.1 Dataset 1
Pima dataset is collected from UCI repository consists of nine features that are
calculated with PySpark approach. Table 12.2 represents the accuracy of different
MLT and their accuracy level: logistic regression (LR), decision tree (DT), random
forest (RF), support vector classifier (SVC) and gradient boosting (GBT). High
accuracy is achieved in RF classifier model with 81 percentage of accuracy and
240 The Internet of medical things

Table 12.2 Pima dataset MLT details

Machine learning Accuracy Test error Time (s)

techniques
LR 0.76 0.24 34
DT 0.71 0.25 56
RF 0.81 0.19 39
SVC 0.76 0.24 40
GBT 0.73 0.27 41

Table 12.3 Raw prediction and probability of LR method

Age Raw prediction Probability

22 1.9 0.87
22 3.93 0.98
22 2.51 0.92
21 3.48 0.97
21 2.31 0.90
25 1.63 0.83
25 1.35 0.79
24 3.43 0.96
23 2.6 0.93
28 0.51 0.62

Table 12.4 Raw prediction and probability of DT method

Age Raw prediction Probability

22 179.0,57.0 0.75
21 43.0,13.0 0.76
24 83.0,1.0 0.98
25 179.0,57.0 0.75
23 43.0,118.0 0.27
28 179.0,57.0 0.75
26 83.0,1.0 0.98
31 179.0,57.0 0.75
41 43.0,118.0 0.26
30 83.0,1.0 0.98

minimum test error of 0.19. The time consumption for running the algorithm is low
in LR model and it is high in GBT model.
Table 12.3 statistics explains the raw prediction and probability of age feature.
The value increases and decreases on the basis of age. It is high at age 25, 22 and
31; only the first 10 rows are listed in the table.
Tables 12.4–12.7 describe the prediction and probability values of different
methods. The value is increasing and decreasing based on the age factor. Based on
 Managing clinical data using machine learning techniques 241

Table 12.5 Raw prediction and probability of RF method

Age Raw prediction Probability

22 16.03 0.80
22 10.14 0.50
21 16.64 0.83
21 15.38 0.76
25 14.96 0.74
25 13.80 0.69
24 13.87 0.69
24 16.75 0.83
23 9.17 0.48
28 14.27 0.74

Table 12.6 Raw prediction and probability of

SVC method

Age Raw prediction Probability

22 16.12 0.80
22 15.1 0.75
21 17.23 0.86
21 16.7 0.83
21 14.6 0.73
24 16.72 0.83
24 16.74 0.81
24 15.06 0.70
25 11.27 0.56
25 16.5 0.82

Table 12.7 Raw prediction and probability of

GBT method

Age Raw prediction Probability

22 1.08 0.89
22 1.36 0.93
21 0.80 0.83
21 1.02 0.88
25 0.54 0.74
24 1.02 0.88
24 1.21 0.91
23 0.08 0.49
28 0.81 0.88
31 1.21 0.91
242 The Internet of medical things

age, the probability value cannot be concluded in different MLT. It is purely based
on the prediction and training results of the algorithm.

12.3.2 Dataset 2
Heart disease data is collected from [15] UCI repository consisting of 303 instances
and 14 features. This data is divided into train and testing and given as input to ML
pipeline process. Table 12.8 summarizes the accuracy of different MLT, where RF
attains high accuracy rate.
Table 12.9 shows the age value of heart disease dataset with raw prediction and
probability. The value of raw prediction is high at the age of 54 and minimum at the
age of 57. Similarly, the probability is maximum at the age of 58 and low at the age
of 57. So, the prediction value is obtained from these statistics value. This table
concludes that if raw prediction value increases, the probability value decreases.
Tables 12.10–12.13 describe clinical dataset 2 manipulated using different
MLT raw predictions and probability value. The value increases and decreases on
the basis of the age factor. In DT method, the values are similar for raw prediction.
Considering RF method at age of 52, the prediction value is maximum and at the
age of 54, the probability value is high. In SVC and GBT, the value is high at the
age of 64. The higher the age value, the prediction value is high.

Table 12.8 Heart disease dataset MLT details

Machine learning Accuracy Test error Time (s)

techniques
LR 0.86 0.3 36
DT 0.88 0.4 32
RF 0.92 0.08 36
SVC 0.90 0.1 30
GBT 0.85 0.15 29

Table 12.9 Raw prediction and probability of LR method

Age Raw prediction Probability

58 4.60 0.99
57 0.85 0.29
54 6.04 0.99
59 3.00 0.95
52 4.09 0.98
51 2.34 0.91
44 2.99 0.95
64 2.5 0.92
41 2.17 0.89
53 1.10 0.75
 Managing clinical data using machine learning techniques 243

Table 12.10 Raw prediction and probability of

DT method

Age Raw prediction Probability

58 [0.0,1.0] 1.0
57 [5.0,17.0] 0.22
54 [42.0,1.0] 0.97
59 [49.0,16.0] 0.75
52 [49.0,16.0] 0.75
51 [49.0,16.0] 0.75
44 [42.0,1.0] 0.97
67 [3.0,53.0] 0.05
65 [3.0,53.0] 0.05
40 [16.0,7.0] 0.69

Table 12.11 Raw prediction and probability of

RF method

Age Raw prediction Probability

58 13.92 0.69
57 7.88 0.39
54 17.77 0.88
59 10.20 0.51
52 16.75 0.83
51 15.56 0.77
62 11.46 0.57
44 17.03 0.85
60 4.77 0.32
40 13.30 0.66

Table 12.12 Raw prediction and probability of

SVC method

Age Raw prediction Probability

58 1.45 0.33
58 1.34 0.43
57 0.10 0.10
54 2.32 0.52
52 1.95 0.27
51 0.17 0.19
44 0.77 0.20
60 0.80 0.27
64 2.44 0.57
62 1.09 0.47
244 The Internet of medical things

Table 12.13 Raw prediction and probability of

GBT method

Age Raw prediction Probability

58 1.34 0.06
54 1.40 0.94
52 0.25 0.62
51 0.59 0.76
62 0.68 0.79
44 0.58 0.76
60 1.05 0.89
64 1.41 0.94
55 0.53 0.74
40 1.01 0.88

Table 12.14 Diabetes disease dataset MLT details

Machine learning Accuracy Test error Time (s)

techniques
LR 0.93 0.3 148
DT 0.92 0.4 130
RF 0.94 0.29 112
SVC 0.93 0.3 142
GBT 0.90 0.5 148

12.3.3 Dataset 3
Diabetes dataset is collected and consists of 100 000 instances. Using feature
selection [19], only 11 features are selected to divide to test and train. This data is
given as input to pipeline MLT observe Table 12.14 results. The accuracy obtained
by RF model is high compared to other models.
Table 12.15 summarizes the raw prediction and probability value based on age.
The prediction value is high at the age of 35 and is minimum at age 65. The
probability value is low at the age of 65.
Tables 12.16–12.19 describe the various raw prediction and probability values
for diabetes dataset. For certain age value, the value is high, which means per-
centage of predicted probability is high.

12.3.4 Dataset 4
Chronic kidney disease dataset is collected from UCI repository [17] and consists
of 400 patient details. Feature consists of a combination of string, integer and
double value. The data is preprocessed using techniques in PySpark. Table 12.20
describes the accuracy and test error values. RF technique obtained high value and
minimum time consumed for classifying the algorithm.
 Managing clinical data using machine learning techniques 245

Table 12.15 Raw prediction and probability of LR method

Age Raw prediction Probability

25 2.12 0.10
35 1.95 0.12
35 2.47 0.07
45 2.13 0.10
55 1.32 0.19
65 0.4 0.39
65 1.14 0.2
45 1.73 0.14
55 1.20 0.23
65 1.12 0.24

Table 12.16 Raw prediction and probability of

DT method

Age Raw prediction Probability

25 [5 066.0,11 742.0] 0.30
35 [5 066.0,11 742.0] 0.30
45 [5 066.0,11 742.0] 0.30
65 [5 066.0,11 742.0] 0.30
65 [5 066.0,11 742.0] 0.30
75 [5 066.0,11 742.0] 0.30
75 [5 066.0,11 742.0] 0.30
75 [5 066.0,11 742.0] 0.30
85 [5 066.0,11 742.0] 0.30
95 [5 066.0,11 742.0] 0.30

Table 12.17 Raw prediction and probability of

RF method

Age Raw prediction Probability

35 8.21 0.41
35 5.76 0.28
45 5.79 0.28
45 7.42 0.37
45 8.44 0.42
55 5.79 0.28
55 7.20 0.36
55 8.31 0.41
55 5.79 0.28
55 5.79 0.28
246 The Internet of medical things

Table 12.18 Raw prediction and probability of

SVC method

Age Raw prediction Probability

25 1.09 0.21
25 1.16 0.26
35 1.04 0.27
35 1.09 0.26
45 1.05 0.21
45 1.03 0.22
45 0.86 0.01
55 0.96 0.1
55 0.95 0.12
55 0.92 0.10

Table 12.19 Raw prediction and probability of

GBT method

Age Raw prediction Probability

25 7.9 0.39
25 5.70 0.28
35 5.76 0.28
35 7.74 0.38
45 5.79 0.28
45 8.26 0.422
55 7.02 0.35
55 8.16 0.40
65 7.20 0.36
65 8.35 0.41

Table 12.20 Chronic kidney disease dataset MLT details

Machine learning Accuracy Test error Time (s)

techniques
LR 0.97 0.03 38
DT 0.96 0.04 39
RF 0.98 0.02 37
SVC 0.95 0.05 36
GBT 0.94 0.06 37

Table 12.21 summarizes the kidney disease raw prediction and probability
value. The raw prediction value is high at the age of 72 and minimum at 63. The
probability value is maximum at the age of 59 and minimum at age 33.
 Managing clinical data using machine learning techniques 247

Table 12.21 Raw prediction and probability of LR method

Age Raw prediction Probability

72 14.85 0.8
63 0.02 0.50
70 4.7 0.90
63 0.02 0.54
59 5.4 0.9
33 3.41 0.03
41 1.2 0.5
72 7.08 0.2
43 5.90 0.2
54 4.97 0.90

Table 12.22 Raw prediction and probability of RF method

Age Raw prediction Probability

53 14.66 0.99
72 15.01 0.99
62 4.89 0.99
30 5.13 0.99
38 1.84 0.86
50 2.4 0.91
39 3.5 0.02
62 3.73 0.02
80 8.36 2.31
46 4.27 0.01

Table 12.23 Raw prediction and probability of DT method

Age Raw prediction Probability

53 [145.0,3.0] 0.97
40 [4.0,98.0] 0.03
44 [145.0,3.0] 0.97
69 [145.0,3.0] 0.97
30 [145.0,3.0] 0.97
55 [145.0,3.0] 0.97
74 [145.0,3.0] 0.97
38 [145.0,3.0] 0.97
60 [145.0,3.0] 0.97
69 [145.0,3.0] 0.97

Tables 12.22–12.25 show the raw prediction and probability value of chronic
disease dataset. The probability and raw prediction based on age factor increases
and decreases.
248 The Internet of medical things

Table 12.24 Raw prediction and probability of SVC method

Age Raw prediction Probability

25 1.09 0.21
25 1.16 0.26
35 1.04 0.27
35 1.09 0.26
45 1.05 0.21
45 1.03 0.22
45 0.86 0.01
55 0.96 0.1
55 0.95 0.12
55 0.92 0.10

Table 12.25 Raw prediction and probability of GBT method

Age Raw prediction Probability

25 7.9 0.39
25 5.70 0.28
35 5.76 0.28
35 7.74 0.38
45 5.79 0.28
45 8.26 0.422
55 7.02 0.35
55 8.16 0.40
65 7.20 0.36
65 8.35 0.41

1.2

0.8
Dataset 1
0.6
Dataset 2
0.4 Dataset 3
0.2 Dataset 4

0
LR DT RF SVC GBT
MLT

Figure 12.3 MLT accuracy calculation of different clinical datasets

 Managing clinical data using machine learning techniques 249

160
140
120
100 Dataset 1
80
Dataset 2
60
40 Dataset 3
20 Dataset 4
0
LR DT RF SVC GBT
MLT

Figure 12.4 Time consumption of ML

Figure 12.3 explains the different datasets with various MLT. Among the
models, RF model that acquires high accuracy is the best for clinical data analysis.
The second highest accuracy is DT. Figure 12.4 describes the time consumed to train
and test the algorithm. The RF acquires minimum time in seconds for large dataset.

12.4 Conclusion
This chapter discusses three different aspects of clinical data. First step is to ana-
lyze the data, integrate the model using pipeline and validate the results. Among the
different benchmark datasets, RF works better in accuracy and time. It also works
best in raw prediction based on age. This helps us to define the usage of MLT and
their scope of clinical data from small to large dataset. Comparison with other
model diagnosis disease better and improve the patient care. For different features,
managing clinical data using MLT gives better results.

References
[1] A. Colubri, T. Silver, T. Fradet, K. Retzepi, B. Fry, and P. Sabeti,
Transforming clinical data into actionable prognosis models: machine-
learning framework and field-deployable app to predict outcome of Ebola
patients. PLoS Negl Trop Dis. 2016; 10(3): e0004549.
[2] S. Dagdanpurev, S. Abe, G. Sun, et al., A novel machine-learning-based
infection screening system via 2013–2017 seasonal influenza patients’ vital
signs as training datasets. J Inf Secur. 2019; 78(5): 409–421.
[3] C. Verdonk, and F. Verdonk, How machine learning could be used in clin-
ical practice during an epidemic. Critical Care, 24(1). doi:10.1186/s13054-
020-02962-y.
[4] S. J. Patel, D. B. Chamberlain, and J. M. Chamberlain. (2018). A machine
learning approach to predicting need for hospitalization for pediatric asthma
exacerbation at the time of Emergency Department Triage. Academic
Emergency Medicine, 2018; 25(12): 1463–1470. doi:10.1111/acem.13655.
250 The Internet of medical things

[5] M. A. Ahmad, and C. Eckert, Interpretable machine learning in healthcare,

BCB’18 Proceeding of the 2018 ACM International Conference on
Bioinformatics, Computational Biology and Health Informatics, August
2018, pp. 559–560.
[6] M. Ghassemi, and T. Naumann, Opportunities in machine learning for
Healthcare. DeepAI. (2018, June 1). Retrieved November 1, 2021, from
https://deepai.org/publication/opportunities-in-machine-learning-for-healthcare.
[7] S. F. Weng, J. Reps, J. Kai, J. M. Garibaldi, and N. Qureshi, Can machine-
learning improve cardiovascular risk prediction using routine clinical data?
PLoS One. 2017; 12(4): e0174944.
[8] A. Callahan, and N. H. Shah, Machine learning in Healthcare. Key Advances
in Clinical Informatics, 2017; 279–291. doi:10.1016/b978-0-12-809523-
2.00019-4.
[9] A. Salcedo-Bernal, and M. P. Villamil-Giraldo, Clinical, Data analysis: an
opportunity to compare machine learning methods, Conference on
ENTERprise Information Systems/International Conference on Project
MANagement/Conference on Health and Social Care Information Systems
and Technologies, Procedia Computer Science 100, 2016.
[10] T. Jiang, Supervised machine learning: a brief primer. Behav Ther. 2020.
[11] A. V. Lebedev, and E. Westman, Random forest ensembles for detection and
prediction of Alzheimer’s disease with a good between-cohort robustness.
Neuroimage Clin. 2014; 6(4): 115–25.
[12] P. H. C. Chen, and Y. Liu, How to develop machine learning models for
healthcare. Nat Mater. 2019.
[13] A. Ashfaq, Predicting clinical outcomes via machine learning on electronic
health records (Licentiate dissertation). 2019. Retrieved from http://urn.kb.
se/resolve?urn=urn:nbn:se:hh:diva-39309.
[14] D. Johnson, PySpark tutorial for beginners: Learn with examples. Guru99.
2021. Retrieved from https://www.guru99.com/pyspark-tutorial.html.
[15] A. Rajkomar, J. Dean, and I. S. Kohane, Machine learning in medicine.
N Engl J Med. 2019; 380(14): 1347–1358.
[16] UCI Machine Learning Repository: Data Set. Retrieved from https://archive.
ics.uci.edu/ml/datasets/heart%C3%BEDisease.
[17] V. Diviya Prabha, and R. Rathipriya, Readmission prediction using hybrid
logistic regression. Innovative Data Communication Technologies and
Application, 2020; 702–709. doi:10.1007/978-3-030-38040-3_80.
[18] P. P. Sengupta, and S. Shrestha, Machine learning for data-driven discovery.
JACC: Cardiovascular Imaging, 2019; 12(4), 690–2. doi:10.1016/j.
jcmg.2018.06.030.
[19] D. A. Clifton, K. E. Niehaus, P. Charlton, and G. W. Colopy. Health infor-
matics via machine learning for the clinical management of patients.
Yearbook of Medical Informatics. 2015; 24(01): 38–43.
 Chapter 13
Use of IoT and mobile technology in virus
outbreak tracking and monitoring
Marimuthu Narayanan Saravana Kumar1, Ravi Bharath2,
Balasubramani Yogeshwaran2, Rajendiran Ranjith2 and
Krishnamoorthy Santhosh Kumar2

13.1 Introduction
The Internet of Things (IoT) has proven useful in the field of electronic-health
(E-Health) management as a network of sensors gathering data both locally and
remotely. The collection of patient vitals and the provision of essential track and
trace services for pandemic management have been made possible thanks to a
combination of body area networks and field monitoring devices. Health data such
as blood pressure (BP), temperature, and heart rate can be collected by locally
based E-Health mechanisms. This data can be saved locally and accessed by a
health-care provider. Local systems can also be used to notify patients when they
need to consult with medical staff to take medicine. Remote-based E-Health is
critical for health-care providers because it allows them to access patients and
patient data from afar. Patient vitals and location may be sent to local or faraway
medical facilities at regular intervals for monitoring purposes. In the event of a
global pandemic, such as the 2019 coronavirus (COVID-19), it is important to
follow social distance guidelines and track and trace patients successfully. These
two factors play an important role in limiting the virus’s global spread. IoT ser-
vices’ ability to provide remote data collection and monitoring of patients in
quarantine has made them a key component in the fight against virus pandemics.
To manage a rapidly spreading respiratory pandemic, health workers and autho-
rities need data. In the case of COVID-19, data can be used to begin the diagnosis
of infection as well as track the spread of the virus in the community. Body tem-
perature, location, and travel history are the most important pieces of information.
These parameters can alert officials as to whether or not more investigation and

1
Department of Electronics and Instrumentation Engineering, Erode Sengunthar Engineering College,
Anna University, Erode, India
2
Department of Biomedical Engineering, Erode Sengunthar Engineering College, Anna University,
Erode, India
252 The Internet of medical things

testing are needed. Initially, health workers relied on manual methods such as
infrared thermometers to measure temperatures and verbal interrogation of people
about their backgrounds and locations. Because of the increased contact with
potentially infected subjects, this posed a risk to health workers. As infection rates
reached the millions, it had also become a more difficult approach.

13.2 IoT in healthcare

IoT applications, in addition to IoT services, deserve more attention. It is worth noting
that services are used to create applications, while applications are used directly by
users and patients. As a result, services are developer-focused, while applications are
user-focused. Various gadgets, wearables, and other health-care devices currently
available on the market are discussed in addition to the application covered in this
section. These products can be thought of as IoT technologies that could lead to a
variety of health-care solutions. The following sections cover a variety of IoT-based
health-care applications, including single- and clustered-condition applications. One of
the most important aspects of the IoT in healthcare is the IoT health-care network, also
known as the IoT network for healthcare (or simply “the INTERNET”). It provides
access to the IoT backbone, as well as the transmission and reception of medical data
and the use of healthcare-specific communications.

13.3 IoT health-care applications

13.3.1 Glucose level sensing
Diabetes is a group of metabolic diseases characterized by persistently high blood
glucose (sugar) levels. Individual patterns of blood glucose changes are revealed by
blood glucose monitoring, which aids in the planning of meals, activities, and
medication times. An m-IoT setup method for real-time noninvasive glucose sen-
sing measuring device. Patients’ sensors are connected to relevant health-care
providers through IPv6 connectivity in this method. The utility model reveals a
transmitting device based on IoT networks for the transmission of collected somatic
data on blood glucose. A generic IoT-based medical acquisition detector that can be
used to monitor glucose levels is another similar innovation.

13.3.2 Electrocardiogram monitoring

Monitoring the electrocardiogram (ECG), or the electrical activity of the heart as recor-
ded by electrocardiography, involves determining the fundamental rhythm and measur-
ing the simple heart rate, as well as diagnosing multifaceted arrhythmias, myocardial
ischemia, and prolonged QT intervals. The IoT is being used in a variety of ways. ECG
monitoring has the potential to provide a wealth of data and can be used to its full
potential. IoT-based ECG monitoring has been debated in a number of studies. Cardiac
function can be detected in real time thanks to the system’s integration of a search
automation method for detecting abnormal data. At the application layer of the IoT
network, there is a comprehensive ECG signal detection algorithm for ECG monitoring.
Use of IoT and mobile technology in virus outbreak tracking and monitoring 253

13.3.3 Blood pressure monitoring

The issue of how a blood pressure (BP) meter kit combined with a near field
communication (NFC)-enabled mobile phone becomes part of IoT-based BP
monitoring is resolved. The communication structure between a health post and the
health center is used to present a motivating scenario in which BP must be peri-
odically controlled remotely. The question of how the Withings BP device works in
conjunction with an Apple mobile computing device is discussed. This device is for
data collection and transmitting the BP data over an IoT network was proposed. A
BP apparatus body and a communication module make up with this device carrying
on with BP tracking and location-aware terminal.

13.3.4 Blood temperature monitoring

Because body temperature is a crucial vital sign in the maintenance of homeostasis,
body temperature monitoring is an important part of health-care services. A body
temperature sensor embedded in the TelosB mote is used to validate the IoT con-
cept, and the results of a typical sample of achieved body temperature variations are
presented, demonstrating the successful operation of the developed m-IoT system.
It is proposed to use an IoT-based temperature measurement system based on a
home gateway. With the assistance of infrared detection, the home gateway trans-
mits the user’s body temperature.
Another temperature monitoring system based on the IoT is proposed. The
radio frequency identification (RFID) module and the module for monitoring body
temperature are the principal system components responsible for temperature
recording and transmission.

13.3.5 Oxygen saturation monitoring

Pulse oximetry is a noninvasive, continuous monitoring system for blood oxygen
saturation. For technology-driven medical health-care applications, the integration of
the IoT with pulse oximetry is beneficial. The potential of IoT-based pulse oximetry is
discussed in a survey of CoAP-based health-care services. The Nonin WristOX2
wearable pulse oximeter is demonstrated in action. This device uses a Bluetooth health
device profile for connectivity, and the sensor connects directly to the Monterey plat-
form. It is proposed to develop an IoT-optimized low-power/low-cost pulse oximeter
for remote patient monitoring. Over an IoT network, this device can be used to con-
tinually monitor the patient’s health. The author describes an integrated pulse oximeter
system for telemedicine applications. A wireless sensor networks (WSN)-enabled
wearable pulse oximeter for health monitoring can be adapted to an IoT network.

13.3.6 Rehabilitation system

Physical medicine and rehabilitation are important branches of medicine because
they can improve and restore functional ability and quality of life in people who
have physical impairments or disabilities. The IoT has the potential to improve
rehabilitation systems by addressing issues such as ageing populations and a
254 The Internet of medical things

shortage of health professionals. For IoT-based smart rehabilitation systems, an

ontology-based automating design method is proposed. This design successfully
illustrates how the IoT can be a useful platform for linking all relevant resources
and providing real-time information interactions. IoT-based technologies have the
potential to create a valuable infrastructure for remote consultation in comprehen-
sive rehabilitation. Many IoT-based rehabilitation systems are available, including
an integrated application system for prisons, hemiplegic patient rehabilitation
training, a smart city medical rehabilitation system, and a language-training system
for children with autism.

13.3.7 Medication management

Noncompliance with prescription poses a serious threat to public health and wastes
a lot of money all over the world. The IoT provides some promising solutions to
this problem. A smart packaging method for medicine boxes is proposed for IoT-
based medication management. This approach involves building a prototype
I2Pack and iMedBox system and testing it in the field. This packaging technique
uses delamination materials that are controlled by wireless communications to
provide controlled sealing. The IoT network is used to present an eHealth service
architecture based on RFID tags for a medication control system. This ubiquitous
medication control system is designed specifically for providing acute lympho-
blastic leukemia solutions, and the prototype implementation is demonstrated here.

13.3.8 Wheelchair management

Many researchers have worked to create fully automated smart wheelchairs for dis-
abled people. The IoT has the potential to speed up the pace of work. On the basis of
IoT technology, a health-care system for wheelchair users is proposed. Wireless body
area networks are integrated with various sensors, functions of which are tailored to
IoT requirements in the design. A peer-to-peer and IoT medical support system is
introduced. This system can detect the status of the wheelchair user and control chair
vibration. The linked wheelchair, developed by Intel’s IoT department, is another
notable example of IoT-based wheelchair development. This progression eventually
demonstrates how commonplace “things” can evolve into data-driven connected
machines. This device can monitor the user’s vitals and collect data on the user’s
surroundings, allowing for the assessment of a location’s accessibility.

13.3.9 Imminent health-care solutions

There are a variety of other portable medical devices available. There is no
explicit demonstration of how those devices can be integrated into IoT networks.
That is, it will only be a matter of time before these devices are equipped with IoT
capabilities. The growing demand for IoT-based services around the world has
resulted in an increase in the number of medical health-care applications, devices,
and cases. Hemoglobin detection, peak expiratory flow, abnormal cellular
development, cancer therapy, eye disorder, skin infection, and remote surgery are
Use of IoT and mobile technology in virus outbreak tracking and monitoring 255

some of the health-care areas where IoT integration appears imminent. The
majority of today’s devices are portable diagnostic devices with traditional
connectivity.

13.3.10 Health-care solutions using smartphones

In recent years, there has been an increase in the number of electronic devices
with a smartphone-controlled sensor, highlighting the rise of smartphones
as a driver of the IoT. To make smartphones a flexible health-care device,
various hardwire and software products have been developed. A thorough
examination of health-care apps for smartphones is given, including a
discussion of apps for patients and general health-care apps, as well as medical
education, training, information search apps, and others (collectively referred to
as auxiliary apps). Furthermore, there are a plethora of new apps that serve
similar functions. Figure 14 shows a classification diagram of auxiliary apps
based on these sources. This figure does not include general health-care apps or
patient-facing apps, which are discussed later in this section. Diagnose and
therapy information can be accessed via diagnostic apps. Reference to a drug
name, indications, dosages, prices, and identifying characteristics are com-
monly provided by apps. Apps for searching biomedical literature databases
make it easier to discover relevant medical information. Tutorials, training,
various surgical demonstrations, color illustrations of various photographs, and
medical books are all common topics in medical education apps. Calculator
apps include a variety of medical formulas and equations that can be used to
calculate different parameters of interest (e.g., the body surface burn percen-
tage). Clinical collaboration apps make it easier for doctors to communicate
inside a hospital. A number of image analysis algorithms for smartphones are
presented that allow for noncontact measurements in health-care applications.
This is a nice (but not exhaustive) list of smartphone apps that provide health-
care solutions. Smartphones are capable of diagnosing and/or tracking the fol-
lowing medical conditions: Asthma, chronic obstructive pulmonary disease,
cystic fibrosis, choking, allergic rhinitis, and nose-related symptoms can all be
detected. In advanced diabetic patients, the respiratory tract, heart rate, BP,
blood oxygen saturation, melanoma, and wound analysis are all examined.
Smartphone health-care apps have a significant advantage in terms of delivering
low-cost solutions, in addition to their pervasive deployment capabilities
and availability for consumers. Many difficulties, such as computational com-
plexity, power consumption, and noisy environments around smartphones,
exist, but they should be simple to overcome. In addition, there are a variety of
health and fitness accessories for smartphones that can assist people in getting
in the best shape possible. Fitbit Flex, for example, is a fitness wristband that
keeps track of steps taken, distance travelled, and calories burned. Existing
commercial health-care products that can be considered as a foundation for IoT
health-care devices are discussed in greater detail in a separate section of
this paper.
256 The Internet of medical things

13.4 Benefits
Smartphone auxiliary
healthcare applications

Johns Hopkins antibiotic guide, Prognosis, Diagnose, 5MCC, 5-Minute

infectious disease consult, MS diagnosis and management, Stanford
guide to anitimicrobial theory, ePocrates ID, Infectious diseases, notes,
Diagnosis apps UpToDate, Pocket medicine infectious diseases, Smart medical-labs,
drug, clac, Palm LabDX, Normal lab values, Lab unit converter, Davis’s
laboratory and diagnostic tests, Video laser level, EyeChart, DizzyFix.

Drug reference Skycape’s RxDrugs, Medical doctor: reference tool, Epocrates,

apps Medscape, SafeMed Pocket, FDA drugs, DrugDoses.net, Lab Values.

Literature PubSearch, PubMed on tap, The/medical search, MD on tap,

search apps askMEDLINE, PICO, Disease associations, Retina medical search.

I-Surgery notebook, Medical diagnosis books, Eponyms, Netter’s atlas

Medical of human anatomy, Diseases dictionary, Netter’s anatomy flash card,
education apps Blausen ear atlas, Oxford handbook of clinical specialties, Dissection,
Cranial nerves, iSilo, iCPR, iResus.

Epocrates MedMath, Calculate by QxMD, MedCalc, Medical calculator,

Calculator apps Calculate, Caddy medical calculator, Archimedes, uBurn Lite,
Softforce’s Antobiotic, Paeds ED, eGFR Calculators.

Clinical
communications Voalte One, Clinical reach, Amcom mobile connect, Practice unite,
Emergency medicine program, Tradassan, mVisum, Vocera.
apps

13.4.1 Simultaneous reporting and monitoring

In the event of a medical emergency such as heart failure, diabetes, or asthma
attacks, real-time monitoring through connected devices will save lives. The IoT
device gathers and transmits health information such as BP, oxygen, and blood
sugar levels, as well as weight and ECGs. These data are stored in the cloud and can
be shared with an approved person, such as a physician, your insurance company, a
participating health firm, or an external consultant, who can access the information
regardless of their location, time, or device.
Use of IoT and mobile technology in virus outbreak tracking and monitoring 257

13.4.2 Data assortment and analysis

When a health-care device’s real-time application sends a large amount of data in a
short period of time, it is difficult to store and handle if cloud access is inaccessible.
Even for health-care providers, manually collecting data from various devices and
sources and analyzing it is a risky proposition. IoT devices can gather, report, and
analyze data in real time, reducing the need for raw data storage. This will all be
done in the cloud, with providers only seeing the final reports with graphs.

13.4.3 Tracking and alerts

In the event of a life-threatening situation, prompt notification is essential. Medical
IoT devices collect vital data and send it to doctors in real time for monitoring, as
well as sending out alerts to people about critical components via mobile apps and
other connected devices. As a result, IoT provides real-time alerting, tracking, and
monitoring, allowing for hands-on treatments, improved accuracy, and appropriate
doctor intervention, as well as improved overall patient care delivery results.
258 The Internet of medical things

13.5 Challenges
13.5.1 Data security and privacy
Data security and privacy are two of the most serious threats posed by IoT. Data is
captured and transmitted in real time by IoT devices. All of these considerations
make the data extremely vulnerable to cybercriminals who can hack into the system
and compromise both patients’ and physicians’ personal health information (PHI).
Cybercriminals may use patient information to produce fake IDs in order to pur-
chase drugs and medical equipment that they will later sell. Hackers will also file a
false insurance claim in the name of a patient.

13.5.2 Cost
You would be surprised to learn that cost factors are included in the challenge parts. I
know most of you are, but the bottom line is that IoT has not yet made healthcare
more accessible to the average person. The rise in health-care costs is a cause for
concern for everyone, particularly in developed countries. The situation has led to the
development of “medical tourism,” in which patients with critical illnesses use
health-care facilities in developed countries for a fraction of the cost. The principle of
IoT in healthcare is a fascinating and promising one.

13.5.3 Data overload and accuracy

Data aggregation is impossible due to the use of various communication protocols and
standards, as previously mentioned. IoT sensors, on the other hand, continue to collect
a large amount of data. The information gathered by IoT devices is used to make
important decisions.
Use of IoT and mobile technology in virus outbreak tracking and monitoring 259

13.6 Use of IoT in virus outbreak and monitoring

13.6.1 Using IoT to dissect an outbreak
IoT would have many more applications during an epidemic because of the
myriad and varied datasets gathered by mobile devices. The IoT should be used to
track down the source of an epidemic. Researchers at Massachusetts Institute of
Technology (MIT) recently used aggregated mobile phone data to track the
spread of dengue fever in Singapore during 2013 and 2014, down to granular
details of short distances and time periods. As a result, using a geographic
information system to overlay IoT mobile data from infected patients will
accomplish two goals. It can aid epidemiologists in their search for patient zero
upstream, and it can aid in the identification of all those who have come into
contact with infected patients and may, thus, be infected downstream.

13.6.2 Using IoT to manage patient care

The scalability of IoT is also useful for keeping track of all the patients who are
high-risk enough to require quarantine but not severe enough to require in-
hospital care. Patients are already checked on a daily basis by health-care pro-
fessionals who go door-to-door. A health-care worker had patients stand on their
apartment balconies so he could fly a drone up and take their temperatures with
an infrared thermometer, according to one report. Patients should have their
temperatures taken and then upload the data to the cloud via their mobile devices
for analysis using IoT. This allows health-care professionals to gather more data
in less time while also reducing the risk of cross-infection among patients. In
addition, IoT will provide relief to the hospital’s overworked staff. The IoT has
already been used to track in-home patients with chronic conditions like
260 The Internet of medical things

hypertension and diabetes. In hospitals, telemetry has been used to track a large
number of patients with limited staff by transmitting biometric measurements
such as heartbeat and BP from wearable, wireless instruments on patients to
central monitoring. IoT should be used to decrease the workload and improve the
productivity of medical personnel while also lowering the risk of infection among
health-care workers.

Doctors

Medical aid Pharna

equipment Companies

Hospitals
Alternaive and
medicine Nursing
Homes

Patient
Care

13.7 Use of mobile apps in healthcare

13.7.1 Mobile report
Patient reports can be mobilized with the assistance of health mobile apps. Not
only can the patient keep track of his or her progress through frequent updates,
but also can the doctor and other hospital officials at any time and from any
location.

13.7.2 Saving human resources

Allocating resources to monitor and retain patient physical data, as well as other
health-care expenditures and earnings, accounts for the majority of the cost. Once
you have got your mobile app up and running, you will be able to save a lot of time
and effort. The same human resources can be put to use in putting other good
thoughts into action in order to better serve the patients.
Use of IoT and mobile technology in virus outbreak tracking and monitoring 261

13.8 Healthcare IoT for virus pandemic management

When IoT technology is used to combat a global virus pandemic, it creates a well-
defined ecosystem of hardware, software, and policies. Based on findings from sur-
veyed studies, this section delves into identifying the components of this particular
health-care ecosystem. The philosophy of using IoT technology to handle a pandemic
results in a unique set of well-integrated components. These elements work together
as part of a virus-fighting ecosystem to stop or slow the spread of a virus.
Other benefits of using IoT to combat a pandemic like COVID-19 include
improved patient management accuracy, lower costs, efficient control, accurate
diagnosis, and the ability to provide superior treatment. As characterized by dif-
ferent research contributions, an ecosystem like this can be divided into the fol-
lowing primary components.

Sensors and Sensors and

End Devices End Devices
Smart City Systems

Networks,Edge Devices,
CloudComputing

Energy Management Transportation, Traffic,

Lighting, Building Parking
Cybersecurity Automation Cybersecurity

Smart City Platforms

A Common System for
Consolidating Data and
Providing Information in
Power and Utilities Context Emergency Management
and Saftey
262 The Internet of medical things

13.9 Evolution of healthcare (pandemic)-based IoT

Several research efforts were invested in “urban intelligence” in the aftermath of
World War II to bring scientific principles into urban activities. As a result, several
of these efforts culminated in the concept of the “smart city.” Indeed, the principle of
urban intelligence outlines three key know-hows: municipal information resources,
science data handling capabilities, and executive power for pandemic preparedness
and response. Skills like data sensing and mining, as well as information integration,
modeling, interpretation, and visualization, have all been used to respond to a pan-
demic. In its most basic form, urban intelligence refers to the use of data science
frameworks and computing methods to solve domain-specific urban challenges. In
recent years, research in the field of the Internet of Medical Things (IoMT) has
suggested a new health-care paradigm known as smart healthcare or intelligent
healthcare (health-care IoT). The earlier research and development activities in
WSN sparked interest in health-care IoT. IoMT primarily deploys Internet-enabled
digitally connected devices with embedded identification, sensing, and data-sharing
capabilities. This would help to close the gap between patients and health-care
providers. Using advances in the management of the health-care system is, without

Community
health
self-reporting and
decentralized
testing
National
communication
Public health
Cloud Mobile phone coordination
Edge
server Global
communication
Internet
Epidemiology
database
Real-time data
update
Radio tower
Connectivity
Geo-spatial
Camera contact
tracina
Boimetric
reader On-board
camera and
sensors

a doubt, the goal of clever healthcare. Intelligent healthcare, for example, uses
wearable devices, a flexible web, and IoT to collect data from people, equipment,
and agencies involved in health-care services and then uses that data to oversee and
respond to health-care needs in a thoughtful manner. Take COVID-19, for example.
By gathering, incorporating, and analyzing exact, appropriate, and high-quality data
in real-time, intelligent healthcare can prevent the virus’s transmission and spread.
To track new COVID-19 cases, intelligent healthcare will gather data through
patient-centered health-based apps. In addition, instead of encounter-based care,
wearable technology (bodyworn-sensors) can be used to provide healthcare to
COVID-19 patients via constant linked care. Additionally, through continuous data
Use of IoT and mobile technology in virus outbreak tracking and monitoring 263

stream and growth, potential COVID-19 hotspots can be proactively identified and
tracked. This makes the virus’s prevention and spread easier. In addition, by com-
bining several data sources, intelligent healthcare will increase community safety.
We examine existing business developments and how they are applied to healthcare
to gain a better understanding of health-based IoT. We have divided the IoT-based
health-care evolution into two categories based on our results. To begin, consider the
use of H-IoT prior to COVID-19. Second, how has the IoT in healthcare react to the
COVID-19 pandemic.

13.10 Increase availability of social networks

Physicians would use mobile technology to consult with other doctors for a second
opinion on their patients. These devices are also beneficial to doctors who work in
different offices in different towns that are relatively far apart. Doctors will use
mobile technology to communicate with patients and coworkers, exchange medical
data, communicate through voice, video, and chat and maintain their workspace.
Doctors will now reach out to more people and assist more patients in more places
thanks to mobile technology.

13.10.1 Improve efficiency of health services

In a variety of ways, mobile technology has aided patients in saving money.
Patients have spent less time in hospitals and have made less visits to their physi-
cians as a result of mobile technology. When doctors offer their personal infor-
mation instead of using the hospital’s phone directory, patients can contact their
doctors much more quickly and effectively. Doctors should speak to their patients
over the phone for a few minutes instead of leaving calls with their receptionist to
follow up if there is not a physical involvement.
Patients’ independence has improved as a result of mobile technology’s ability
to help them improve their health. Patients have benefited from Equated Monthly
Installment’s (EMIs) continuous monitoring of their health conditions. This device
has enabled patients to take responsibility for their health and reduce their reliance
on doctors and nurses. The EMIs have given patients information about their
symptoms and what they need to do about their newly diagnosed disease. EMIs,
according to the author, provide information on barriers that prevent patients from
quitting smoking and losing weight by showing progress after the device has been
used daily. Having a device to keep track of the improvements that need to be made
to achieve one’s personal health goals instead of visiting physicians, nurses, and/or
nutritionists is a fantastic benefit for patients.

13.10.2 Improve patients health condition

The primary reason why mobile technology has become so prevalent in hospitals is
that it has improved the lives of many patients. According to Avancha (2012),
mobile technology has enabled doctors to keep track of their patients more reg-
ularly, increase the quality of health care, and make it easier for patients to take
264 The Internet of medical things

care of their own. Doctors have been able to better serve their patients in a timely
fashion thanks to mobile technology. According to Heron (2010), equated monthly
installment (EMI) has aided in the treatment of a variety of disorders, including
obsessive-compulsive disorder, generalized anxiety disorder, panic disorder, and
social phobia. There are a slew of other ailments for which mobile technology has
made a difference in people’s lives.

13.10.3 Enhances physician efficiency

Physicians who are well-functioning are the foundation of a well-functioning
health-care system. However, it is a high-stress position. Approximately 60% of
emergency medicine physicians, according to the American Medical Association,
are burnt out. Physicians often lament the amount of time they spend on data entry
and other administrative activities. Only 27% of their time is spent caring for
patients. To combat this issue, service providers are looking for ways to make the
medical profession less stressful. Doctors are using phones and tablets to fix pro-
blems quicker and with less stress.
Physicians should record patient history on mobile devices with zero errors. These
devices give users easier access to the most up-to-date drug information, allowing them
to make better choices. They automate a lot of paperwork, freeing up time for doctors. It
also means that doctors now have access to all of their patients’ information, allowing
them to spend less time delivering the same level of care. Overall, the use of mobile
devices in healthcare has increased physician productivity and improved patient care.

13.11 Data privacy and security is a significant concern

Data privacy and security have long been a source of concern for health-care
institutions. This was the case before smartphones and other devices, such as
notebooks, tablets, and portable hard drives, allowed information to freely travel
outside their facility walls. If a smartphone is lost or stolen and does not have
adequate security, the data stored on it and available via apps can be viewed and
potentially downloaded. A situation like this might have serious legal and financial
ramifications for a company and its clinicians.

13.11.1 Lack of information control

Many hospitals have expressed fears about the lack of control over how an individual
would gain access to some information and who may possibly gain access to other
people’s information. Technology has progressed to the point where someone might
potentially gather significant quantities of data without anyone’s knowledge. Many
people are concerned about the confidentiality of their personal information as a
result of this problem. Passwords restrict the confidentiality of information on mobile
technology. Anyone might accidentally tap into someone’s phone, which could cause
a slew of issues. The primary problem with mobile technology is that it is difficult for
health administrators to use it, and the hospital must adhere to numerous privacy
Use of IoT and mobile technology in virus outbreak tracking and monitoring 265

regulations. Software is used by 65% of businesses to block websites. However, 35%

of businesses are not adhering to medical privacy regulations, which may cause a
slew of issues for patients and their families. Patients may not want certain people in
their lives, such as their boss, to know about their health conditions because it may
put them at a disadvantage at work, or family members who may profit from the
information that the patient does not want to share. Strong regulations, such as the
Health Insurance Portability and Accountability Act (HIPAA), mandate patient–
doctor secrecy. HIPAA also stresses protecting PHI from health-care providers and
mandates that health information in electronic form be protected, monitored, and
controlled through organizational access protection, monitoring, and control.

13.11.2 Digital divide among patients

Hospitals should be available to all patients, regardless of their socioeconomic
status. There are also people who do not have a smartphone or even a computer to
contact via e-mail for more frequent updates. Hospitals with a high proportion of
poor patients lag behind in creating digital technologies, and administrators find it
impossible to change their systems to become more digital. This problem of the
digital divide has the potential to result in inequalities in healthcare.

13.12 Conclusion
We suggest a model for IoT-based health-care systems in this chapter, which can be
used for both general systems and systems that monitor special conditions. Then,
for each component of the proposed model, we submitted a detailed and systematic
overview of the state-of-the-art works. Several nonintrusive, wearable sensors were
demonstrated and evaluated, with a focus on those that monitor vital signs, BP, and
blood oxygen levels. The suitability of short-range and long-range communication
requirements for health-care applications was then compared. For short-range and
long-range communications in healthcare, bilateral lower extremity and Narrow
Band-Internet of Things (NB-IoT) emerged as the most suitable standards. Recent
cloud-based data storage research was introduced, demonstrating that the cloud is
the best option for storing and coordinating big data in healthcare. Several studies
have also found that data processing in the cloud is much better than data proces-
sing on wearable devices with their limited resources. The most important dis-
advantage of using the cloud is that it adds security risks; as a result, we introduced
several works aimed at enhancing cloud security. Access control policies and
encryption were discovered to substantially improve security, but no known stan-
dard is suitable for immediate implementation in a wearable, IoT-based health-care
system. We found several important areas for future study based on our analysis of
state-of-the-art technologies in the fields of wearable sensors, communication
standards, and cloud technology. Machine learning and the development of a secure
but lightweight encryption scheme for cloud storage are the two areas where
researchers looking to make substantial changes in the field of IoT-based healthcare
have the most opportunities.
266 The Internet of medical things

Further reading
[1] M.U. Ahmed, S. Begum, and J.-B. Fasquel (Eds.) 4th International
Conference, HealthyIoT 2017, Angers, France, 2017, pp. 3–9.
[2] K. Govinda Contemporary Applications of Mobile Computing in Healthcare
Settings, VIT University, Vellore, India, 2018, pp. 51–57. Copyright: ’
2018.
[3] A.K. Singh, R.S. Singh, A.K. Pandey, S. Udmale, and A. Chaudhary ISBN:
9780128214725, 11th November 2020, pp. 36–59.
[4] C. Chakraborty, A. Banerjee, A. Kolekar, H. Maheshkumar, and B.
Chakraborty Internet of Things for Healthcare Technologies, March 2020
Pages: 21–43. ISBN: 978-981-15-4111-7.
[5] A. Gantait, J. Patra, and A. Mukherjee Defining Your IoT Governance
Practices, January 19, 2018. Updated January 20, 2018.
[6] P. Sharma How Edge Computing in Healthcare Is Transforming IoT
Implementation. Apc.com, December 12, 2017.
[7] Edge Computing for IoT in Buildings. Navigant Research, Q4 2018.
[8] Global Smart Healthcare Products Market Will Reach USD 66.7 Billion by
2024, Zion Research, January 2019.
[9] Gartner, IT Glossary, Internet of Things http://www.gartner.com/it-glossary/
internet-of-things/.
[10] Gartner, Press release, 2013, online at http://www.gartner.com/newsroom/id/
2636073.
[11] ITU Internet Reports, The Internet of Things, November 2005 http://www.
itu.int/osg/spu/publications/internetofthings/InternetofThings_summary.pdf.
[12] IERC – European Research Cluster on the Internet of Things, “Internet of
Things – Pan European Research and Innovation Vision”, October, 2011.
[13] L. Adori, A. Iera, and G. Morabito The Internet of Things: A Survey,
Computer Networks, 54 (2010) 2787–2805.
[14] H. Jun-Wei, Y. Shouyi, L. Leibo, Z. Zhen, and W. Shaojun A Crop
Monitoring System Based on Wireless Sensor Network, Procedia
Environmental Sciences, 11 (2011) 558–565.
[15] A.M. Vilamovska, E. Hattziandreu, R. Schindler, C. Van Oranje, H.
DeVries, and J. Krapelse RFID Application in Healthcare – Scoping and
Identifying Areas for RFID Deployment in Healthcare Delivery, RAND,
Europe, 2009.
[16] P. Pande Internet of Things – A Future of Internet: A Survey, International
Journal of Advance Research in Computer Science and Management
Studies, 2(2), 2014.
[17] Internet of Things: From Research and Innovation to Market Deployment –
IERC 2014. http://www.internet-of-things-research.eu/pdf/IERC_Cluster_
Book_2014_Ch.3_SRIA_WEB.pdf.
[18] Proteus, Digital health feedback system, http://www.proteus.com/technol-
ogy/digital-health-feedback-system.
Use of IoT and mobile technology in virus outbreak tracking and monitoring 267

[19] Proteus, Digital Medicine, http://www.proteus.com/future-products/digital-

medicines/.
[20] D. Christin, A. Reinhardt, P. Mogre, and R. Steinmed “Wireless Sensor
Networks and the Internet of Things: Selected Challenges,” in Proceedings
of the 8th GI/ITG KuVS Fachgespräch “Drahtlose Sensornede”, 2009,
pp. 31–33.
[21] C. Li, A. Raghunathan, and N. Jha “Hijacking an Insulin Pump: Security
Attacks and Defenses for a Diabetes Therapy System,” in IEEE 13th
International Conference on e-Health Networking, Applications and
Services, Columbia, MO, 2011, pp. 150–156.
[22] Wikipedia, Personalized medicine http://en.wikipedia.org/wiki/Personalized_
medicine.
[23] IEEE Newsletter, P. Desikan, R. Khare, J. Srivastava, R. Kaplan, J. Ghosh,
L. Liu, and V. Gopal Predictive Modeling in Healthcare: Challenges and
Opportunities http://lifesciences.ieee.org/publications/newsletter/november-
2013/439-predictive-modeling-in-healthcare-challenges-and-opportunities.
[24] M. Abo-Zahhad, S. M. Ahmed, and O. Elnahas A Wireless Emergency
Telemedicine System for Patients Monitoring and Diagnosis. https://www.
who.int/gho/mortality_burden_disease/life_tables/situation_trends_text/
en/24.
[25] S. Gotadki, R. Mohan, M. Attarwala, and M.P. Gajare “Intelligent
Ambulance”.
[26] C. Li, X. Hu, and L. Zhang “The IoT-Based Heart Disease Monitoring
System for Pervasive Healthcare Service,” in International Conference on
Knowledge Based and Intelligent Information and Engineering Systems,
KES2017, Marseille, France, 6–8 September 2017.
[27] L. Yu “Smart Hospital based on Internet of Things”.
[28] D.S. Abdul Minaam and M. Abd-ELfattah Smart Drugs: Improving
Healthcare Using Smart Pill Box for Medicine Reminder and Monitoring
System, Future Computing and Informatics Journal, 3(2018) 443–456.
[29] P. Deepika Mathuvanthi, V. Suresh, and Ch. Pradeep IoT Powered Wearable
to Assist Individuals Facing Depression Symptoms, International Research
Journal of Engineering and Technology (IRJET), 6 (2019) 1676.
[30] A. Haleem, M. Javaid, and I.H. Khan Internet of Things (IoT) Applications
in Orthopaedics, Journal of Clinical Orthopaedics and Trauma (2019), doi:
https://doi.org/10.1016/j.jcot.2019.07.003.
[31] M. AboZahhad, S.M. Ahmed, and O. Elnahas A Wireless Emergency
Telemedicine System for Patients Monitoring and Diagnosis, International
Journal of Telemedicine and Applications, 2014 (2014) 380787.
 Chapter 14
Video-based solutions for newborn monitoring
Veronica Mattioli1, Davide Alinovi1,*, Francesco Pisani2,
Gianluigi Ferrari1 and Riccardo Raheli1

Efficient monitoring of vital signs is a fundamental tool in disease prevention and

medical diagnostics. Main physiological parameters to monitor are not only heart
rate, blood pressure, respiratory rate and body temperature, but motion analysis
may also provide essential information about the clinical status of a patient. Very
specific pathological movements can indeed be signs of important or potentially
threatening disorders. Besides being almost exclusively performed in hospital set-
tings, conventional monitoring often requires a contact with the body of the patient
that makes traditional systems possibly invasive and uncomfortable, especially if
applied on newborns. To make home care more accessible and comfortable, novel
methods for remote and contactless monitoring have been developed in the recent
years. Among others, appealing solutions that have received recent research
attention are based on video processing techniques that allow to capture and ana-
lyze the movements of a patient in a contactless fashion.

14.1 Introduction

Early diagnosis of neonatal disorders may be crucial for timely intervention and
treatment. As some rare but potentially harmful diseases in newborns that can man-
ifest themselves with clinical symptoms, such as seizures and apneas, affect the
movements of the patient, movement monitoring and analysis may be an effective
diagnostic tool. In particular, some types of seizures can be characterized by jerky
periodic movements of one or more body parts, usually limbs and head. On the other
hand, apneas are associated with the absence of periodic breathing movements [1].
A seizure can be defined as an age-dependent clinical event characterized by a
neurological dysfunction caused by paroxysmal alterations of neurological, beha-
vioral and/or automatic functions [2]. One of the most common outward effects is
an uncontrolled shaking due to involuntary and rapidly contraction and relaxation

1
Department of Engineering and Architecture, University of Parma, Parma, Italy
2
Department of Medicine and Surgery, University of Parma, Parma, Italy
*
Deceased on 16 September 2020
270 The Internet of medical things

of one or more muscle groups. Preterm and at-term newborns are more likely to
suffer from a seizure within 28 days after birth or 44 weeks of conceptional age,
respectively [2]. The estimated incidence is 2:6% for overall newborns, 2:0% for
term neonates, 11:1% for preterm neonates and 13:5% for infants weighing less
than 2 500 g [3]. Hypoxic–ischemic encephalopathy and stroke are not only two of
the more frequent etiologies, but brain malformation and infection can also be
triggers [4]. As reported in [4,5], several classifications have been proposed, but
usually four main types of clinical manifestations are considered indicative of
neonatal seizures: subtle, clonic, tonic and myoclonic [2]. Each clinical type of
seizure is characterized by distinguishable features and requires focused analysis
and diagnostic approach. Clonic seizures, for instance, are associated with rhythmic
and slow movements.
On the other hand, apneas can be defined as sudden interruptions of the
respiratory airflow. In newborns, these episodes are considered to be significant if
lasting longer than 20 s or less if associated with other symptoms, i.e., bradycardia
and cyanosis [6]. As reported in [1], among the main causes of neonatal apneas, we
recall seizures, cerebrovascular events [7] and congenital disorders, such as con-
genital central hypoventilation syndrome (CCHS) [8,9]. CCHS, in particular, is a rare
life-threatening disease caused by a defect in the PHOX2B homeobox gene [9]. It
mainly occurs during sleep and is responsible of alveolar hypoventilation. It is
usually associated with cyanosis, apnea or cardiorespiratory arrest [10]. Finally, three
main categories of apneas can be identified, i.e., central, obstructive and mixed,
according to the presence or the lack of an obstruction of the upper airway [6].
Due to the severity of these neonatal disorders, early treatments are needed to
prevent life-threatening episodes as well as lifelong consequences. To this purpose,
efficient monitoring tools must be deployed. The investigation of modern mon-
itoring systems based on video processing solutions can be considered a promising
and effective alternative to conventional equipment. Motion analysis plays a key
role to detect anomalous movements related to the aforementioned disorders and
will be discussed in the next sections.

14.2 Vital signs monitoring

Conventional systems for monitoring vital signs are often intrusive and not
suitable for home care. The main standard tool for the diagnosis of sleep-related
disorders, such as seizures and apneas, is the polysomnogram that allows to record
the sleep of a patient and includes several monitoring systems, i.e., electro-
encephalogram (EEG), electrooculogram, electromyogram and electrocardiogram
[11]. Each of these measurement techniques requires wired sensors to be directly
attached to different body parts of the patient. In particular, electrodes placed on the
scalp, near the eyes, under the chin and on the chest are employed to record brain
activity, eyes movements, muscle activity and heart rate, respectively. Additional
information is acquired through elastic belt sensors placed around the chest, nasal
flow meter and pulse oximeter to measure the amount of effort to breathe, the
 Video-based solutions for newborn monitoring 271

airflow and the oxygen saturation of the blood, respectively [11]. Cameras can also
be employed for simultaneous traditional monitoring. A schematic overview of the
polysomnogram test is shown in Figure 14.1(a) and an example of recorded data is
reported in Figure 14.1(b) where the first four traces are the EEG channels, and the
subsequent traces are, from top to bottom, snoring noise, nasal flow, thoracic
movements and oxygen saturation. The abnormal breathing pattern is characterized
by recurrence of central apneas (closed boxes), in the absence of airway obstruction
and snoring. Central apneas determine severe oxygen desaturations.
Besides being expensive and moderately invasive, especially for newborns,
these techniques are almost exclusively deployed in clinical settings and require
trained medical staff who may not be available full time. To make home care more
accessible, various monitoring systems have been developed, e.g., smart bed [12]
and wearable-sensor-based systems [13], but they still require contact with the body
of the patient. Contactless solutions, on the other hand, may be devised for the
automatic detection of anomalous activities potentially related to neonatal dis-
orders. To this purpose, digital cameras can be used to frame the movements of a
patient to be analyzed through proper video processing algorithms. The integration
of these novel approaches allows to enhance both hospital and home constant
monitoring by providing low-cost preliminary alert signals to be possibly further
investigated by conventional diagnostic tools, i.e., the EEG.

14.3 Video processing systems for neonatal disorder

detection
Early work on video-based solutions for newborn monitoring was mainly focused on
seizure detection and was based on motion extraction algorithms such as optical flow
[14] and block matching [15]. Neural networks for anomalous event detection and
motion classification were also investigated. The implementation of all these meth-
ods may be complex, expensive and not suitable for real-time monitoring. However,
in [1] a fast and reliable approach for the real-time analysis of newborns’ movements
is proposed on the basis of preliminary contributions [16–18]. This method is based
on the extraction of motion signals acquired with single or multiple cameras and
ultimately relies on the well-known maximum likelihood (ML) estimation criterion
[19]. Besides being presented for clonic seizures and apneas detection, it is valid for
any disorder characterized by the presence or the absence of periodic movements.
In Sections 4.3.1 and 14.3.2, the motion estimation algorithm presented in [1]
will be briefly introduced for single- and multiple-sensor analysis, respectively. The
performance in the latter case is improved by the different viewing angles that allow
to capture movements that may be occluded for a single camera, hence undetectable.
In particular, models to describe motion signals acquired from properly preprocessed
video sequences are discussed. The specific procedures for the extraction of motion
information related to seizures and apneas will be presented in Sections 14.4 and
14.5, respectively. Since the disorders under investigation are characterized by the
presence or absence of periodic movements, motion signals will be modeled as
272 The Internet of medical things

Sensors to
measure brain activity,
eye movements and
muscle activity
Sensors to
measure air flow
and breath

Sensor to
measure oxygen
in blood

(a)

(b)

Figure 14.1 Polysomnography during wakefulness: (a) schematic overview and

(b) example of recorded data. Image by Francesca and Andrea
Pisani
 Video-based solutions for newborn monitoring 273

periodic signals, where the fundamental frequency represents the main unknown
parameter to be estimated.
In the following analysis, we will consider video sequences with sampling
period, T , where frames have dimension W
H and are sampled at time instants
iT , i being the frame number.

14.3.1 Single sensor

 be the average luminance signal extracted at frame i of a considered video
Let L½i
sequence. We define it as
XW X H
 ¼ 1
L½i I½x; y; i (14.1)
WH x¼1 y¼1

Where I½x; y; i represents the ½x; y entry of matrix I½i that describes the ith
frame after a proper processing procedure. The motion signal in (14.1) can be
modeled as
 ¼ c þ Acosð2p f0 iT þ jÞ þ n½i
L½i (14.2)
Where c is a continuous component and n½i are samples of independent
identically distributed zero-mean Gaussian noise. The unknown parameters A, f0
and j represent the amplitude, phase and frequency, respectively, of the periodic
signal and may be collected in a vector q ¼ ½A; f0 ; f. The ML approach can now be
exploited to estimate the vector q. In particular, observing a window of N frames
and following standard methods described in [20], an estimator of the fundamental
frequency can be obtained and expressed as
 2
XN 1 
b   j2pfiT 
f 0 ¼ argmaxf  L½ie  : (14.3)
 i¼0 

Similarly, an expression of the amplitude estimator can be written as

 
b 2 NX
1
 j2p b


A¼  L½ne f 0 iT
: (14.4)
N  n¼0 

The presence of a significant periodic component can be finally declared if the

following constraint is verified:

b >h
NA
2
(14.5)
Where the value of the threshold h may be determined by trial and error.

14.3.2 Multiple sensors

The method described in Section 14.3.1 can be extended to multiple sensors to
achieve better performance. Multi-camera systems can indeed detect movements
274 The Internet of medical things

that may be occluded for a single camera. Considering S sensors, a set of motion
signals is defined as in (14.1):
XW X H
 s ½i ¼ 1
L Is ½x; y; i; s ¼ 1; 2; . . .; S: (14.6)
WH x¼1 y¼1

Where the processed ith frame for the sth sensor is described by matrix
Is ½x; y; i. The model in (14.2) can be generalized as
 s ½i ¼ cs þ As cosð2p f0 iT þ js Þ þ ns ½i
L (14.7)
Where the sampling period T and the fundamental frequency f0 are assumed to
be identical for each capturing device; where present, the subscript s refers to the
sth sensor. Following the same procedure of the single-sensor analysis and
exploiting now data fusion techniques to combine data acquired by different sen-
sors, an estimator of the fundamental frequency can now be formulated as
 2
X S NX1 
b   s ½iej2pfiT  :
f 0 ¼ argmaxf  L (14.8)

s¼1 i¼0


Likewise, assuming that different values of amplitude are associated to each

sensor, a set of amplitude estimators is:
 
XN 1 
bs ¼  2   s ½ne j2pb 
A L f 0 iT
; s ¼ 1; 2; . . .; S: (14.9)
N  n¼0 

Finally, the constraint to be satisfied in order to detect a significant periodic

component is now

NX S
b 2 > h:
A (14.10)
S s¼1

To improve the performance of both single- and multi-sensor analyses, interlaced

windows can be considered, as the selected detection algorithm may indeed fail when
pathological movements manifest across two consecutive disjoint windows [16].

14.4 Seizure detection

In order to extract the average motion signals defined in (14.1) and (14.6), each frame
of a considered video sequence needs to be properly processed. A schematic overview
of the preprocessing algorithm exploited in [1,16] for seizure detection is presented in
Figure 14.2, where four phases are highlighted: grayscale conversion, difference fil-
tering, binarization and erosion. A generic red, green and blue (RGB) video sequence
X½i is considered the input of the processing system and initially converted to gray
scale. The difference of frames (DoF) is then performed on consecutive frames as a
 Video-based solutions for newborn monitoring 275

Video DoF
input
z –1
X[i] D[i] Binarization

Spatial
L[i] I[i] B[i]
average
Erosion

Figure 14.2 Seizure detection preprocessing algorithm

(a) (b) (c) (d)

Figure 14.3 Results of each processing step: (a) gray-scale, (b) difference
filtering, (c) binarization and (d) erosion

basic image filtering operation and the result is threshold to obtain a binary mask,
where white pixels correspond to foreground regions. Finally, the erosion morpholo-
gical operation [20] is implemented to reduce noise as discussed in [16]. Examples of
frames at each processing step are shown in Figure 14.3. Eventually, an example of
periodic movements induced by a clonic seizure and extracted according the procedure
illustrated in Figure 14.2 is shown in Figure 14.4, where the extracted average motion
signal is plotted against the frame number along with a corresponding EEG signal [16].
The two signals exhibit a comparable periodicity.

14.4.1 Performance in seizure detection

The performance of the described detection method has been evaluated in terms of
sensitivity and specificity over n tests, respectively, defined as
nTP
a¼ (14.11)
nTP þ nFN
nTN
b¼ (14.12)
nTN þ nFP
where nTP , nTN , nFP and nFN are the numbers of true positives, true negatives, false
positives and false negatives in the considered sequence. In particular, positives and
negatives are classified when a seizure is detected or undetected, respectively.
Performance of single- and multi-sensor systems have been investigated and the
276 The Internet of medical things

Average motion [pixels] EEG Fp2 C4 signal

Average
motion
signal

i [Frame number]

Figure 14.4 Average motion signal [16]

1 1
D = 0.14
0.9 Camera 1
0.8 0.9 Camera 2
Camera 3
0.7
Camera 1 and 2
0.6 0.8 Camera 1 and 3
AUC = 0.95
0.5 Camera 2 and 3
α

Camera 1,2 and 3

0.4 0.7
0.3
0.2 0.6
0.1
0 0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5
(a) 1−β (b) 1−β

Figure 14.5 Performance analysis for (a) single RGB camera (ROC curve); (b)
different RGB camera configurations [1]

results obtained using a single RGB sensor are illustrated in Figure 14.5(a), where
the receiver operating characteristic (ROC) [21] curve is plotted as a function of a
and 1  b for various values of the threshold h. We recall that an optimal predictor
is characterized by a ¼ 1 and b ¼ 1, i.e., all seizures are correctly detected when
present, and an area under the curve (AUC) is equal to 1. In the presented example,
the minimum Euclidean distance D from the ideal configuration is 0.14 and the
AUC is 0.95, which indicates high reliability [21]. In Figure 14.5(b), sensitivity and
specificity values are plotted for optimal values of h and different RGB cameras
configurations. The best performance is achieved when all three sensors are
employed, i.e., S ¼ 3 in (14.6)–(14.10). Depth sensors could also be employed to
better distinguish pathological movements from background noise or random
movements [1].
 Video-based solutions for newborn monitoring 277

14.5 Apnea detection

Breathing-related movements are often subtle and difficult to detect, especially for
newborns. To make the detection algorithm efficient also in the presence of small
movements, a motion magnification algorithm can be applied to amplify the con-
sidered motion signals. In particular, the Eulerian video magnification (EVM)
method presented in [22] can be exploited and its schematic representation is
illustrated in Figure 14.6, where the input signal X½i is processed through four
main phases. Initially, each frame of the considered video sequence is decomposed
into different spatial frequency bands by spatial decomposition. The obtained out-
puts are then filtered through a pixel-wise temporal operation and the frequency
bands of interest are extracted. Multiplications by proper gains are now performed
to amplify the filtered signals and a video frame reconstruction is finally imple-
mented to obtain a new output signal where small changes at the input are
enhanced. The motion extraction algorithm shown in Figure 14.2 can be applied
after the EVM processing.
An example of an extracted motion signal is plotted in Figure 14.7(a), where
the normal periodic behavior of a respiratory signal can be easily observed. The
corresponding periodogram is shown in Figure 14.7(b), where the estimated

P0
P0 ϒ0 α0
Temporal BP-IIR filter
(pixel-by-pixel)
decomposition

reconstruction
P1
Spatial

Spatial

P1 ϒ1 α1
X[i]

Video Video
input input

PL–1 ϒL–1 αL–1 PL–1

Figure 14.6 EVM algorithm [18]

1 0.03
Average luminance signal

0.025
0.8
Periodogram of L

0.02 ^
0.6 f : estimated frequency
0.015
0.4
0.01
0.2 0.005

0 0
0 0.5 1 1.5 2 2.5 3
0 100 200 300 400 500 600 700
Frame (b) Frequency [Hz]
(a)

Figure 14.7 (a) Periodic motion signal example and (b) periodogram
278 The Internet of medical things

0.8
Average luminance signal

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14 16 18
Time [s]

Figure 14.8 Anomalous motion signal related to an apnea event in a newborn

×10−3
400 4

Average luminance signal

300
Pneumogram [μV]

200 2
100
0
0
–100
–200 Pneumogram
Motion signal –2
–300
0 5 10 15 20
Time [s]

Figure 14.9 Extracted motion signal and correspondent pneumographic signal

fundamental frequency is highlighted at the peak of the function. On the other hand,
an example of an anomalous motion signal is illustrated in Figure 14.18, where the
sudden interruption of the respiration caused by an apnea episode is visible in the
flat central part of the plot.
Finally, for the sake of comparison, an instance of an extracted motion signal is
shown in Figure 14.9 along with the equivalent signal obtained from a pneumograph,
where every period of the pneumographic signal corresponds to a complete respira-
tory act of the patient. Considering that a respiratory act is composed by two phases,
i.e., inhalation and exhalation, a good match of the two signals can be observed.
A number of improvements are possible with reference to the method descri-
bed in Figure 14.6. Among them, we mention [23] where the computationally
intensive reconstruction of the video stream is avoided and the sublevel signals are
directly combined in a multidimensional effective estimator. Direct application of
 Video-based solutions for newborn monitoring 279

×10–3
1

Average luminance signal

8

6 TD = 30 s
TD = 20 s
TD = 10 s
4 TD = 0 s

2
20 s

0
0 10 20 30 40 50 60
Apnea onset Time [s] Observation window
Negative test
Positive test

Figure 14.10 Detection of a long apnea episode

multidimensional ML estimator to the video stream was also investigated and

showed very good performance, provided proper processing is performed to select
a suitable region of interest [24]. We also remark that these methods could be
applied to seizure detection as well. However, the subtle respiration movements
specifically require these more sensitive (and complex) solutions.

14.5.1 Performance in apnea detection

To evaluate the performance of the apnea detection algorithm, sensitivity and
specificity coefficients are still considered but slightly differently defined as
TTP
a¼ (14.13)
TTP þ TFN
TTN
b¼ (14.14)
TTN þ TFP
where TTP , TTN , TFP and TFN represent now the total duration of time intervals when
an apnea episode is correctly detected (time true positives), correctly undetected
(time true negatives), incorrectly detected (time false positives) and incorrectly
undetected (time false negatives), respectively. An interval of tolerance, i.e., tol-
erance delay (TD), is allowed to declare that an apnea episode is detected. As an
illustrative example, the motion signal related to a long apnea episode is shown in
Figure 14.10, where half interlaced observation windows are highlighted in the
upper part of the plot and different TD values are considered. The apnea detection
fails for TD ¼ 0 s, while it succeeds for TD ¼ 10; 20 and 30 s.

14.6 Conclusion

In this chapter, novel techniques for newborn monitoring based on video processing
solutions have been proposed. Considering that disorders such as seizures and apneas
280 The Internet of medical things

are characterized by specific pathological movements, proper algorithms for motion

analysis have been presented. In particular, as clonic seizures trigger jerky move-
ments of some human body parts and apneas cause sudden interruptions of the
rhythmical respiration movements, periodicity is the fundamental parameter to be
investigated. Motion signals extracted from video sequences can indeed be modeled
as periodic signals, where the fundamental frequency can be ultimately estimated by
standard probability theory techniques. The main goal is the integration of reliable,
noninvasive and contactless systems with conventional equipment to provide clinical
support to enhance early diagnosis of potentially life-threatening diseases.

References
[1] Cattani L, Alinovi D, Ferrari G, et al. Monitoring infants by automatic video
processing: A unified approach to motion analysis. Computers in Biology
and Medicine. 2017; 80: 158–165.
[2] Volpe JJ. Neurology of the Newborn. 5th ed. Philadelphia, PA, USA:
Saunders Elsevier; 2008.
[3] Ronen GM, Penney S, and Andrews W. The epidemiology of clinical neo-
natal seizures in Newfoundland: A population-based study. The Journal of
Pediatrics. 1999; 134(1): 71–75.
[4] Pellegrin S, Munoz FM, Padula M, et al. Neonatal seizures: Case definition
and guidelines for data collection, analysis, and presentation of immuniza-
tion safety data. Vaccine. 2019; 37(52): 7596–7609.
[5] Riviello JJ. Classification of seizures and epilepsy. Current Neurology and
Neuroscience Reports. 2003; 3: 325–331.
[6] Mishra S, Agarwal R, Jeevasankar M, et al. Apnea in the newborn. Indian
Journal of Pediatrics. 2008; 75: 57–71.
[7] Tramonte J and Goodkin H. Temporal lobe hemorrhage in the full-term neonate
presenting as apneic seizures. Journal of Perinatology. 2004; 24: 726–729.
[8] Rand CM, Patwari PP, Carroll MS, et al. Congenital central hypoventilation
syndrome and sudden infant death syndrome: disorders of autonomic reg-
ulation. Seminars in Pediatric Neurology. 2013; 20(1): 44–55. Pediatric
Autonomic Disorders.
[9] Healy F and Marcus CL. Congenital central hypoventilation syndrome in
children. Paediatric Respiratory Reviews. 2011; 12(4): 253–263.
[10] Cielo CM and Marcus CL. Central hypoventilation syndromes. Sleep
Medicine Clinics. 2014; 9(1): 105–118.
[11] Martin RJ, Block AJ, Cohn MA, et al. Indications and standards for cardi-
opulmonary sleep studies. Sleep. 1985; 8(4): 371–379.
[12] Spillman Jr WB, Mayer M, Bennett J, et al. A smart bed for non-intrusive
monitoring of patient physiological factors. Measurement Science and
Technology. 2004; 15(8): 1614–1620.
[13] Pantelopoulos A and Bourbakis NG. A survey on wearable sensor-based
systems for health monitoring and prognosis. IEEE Transactions on Systems,
 Video-based solutions for newborn monitoring 281

Man, and Cybernetics, Part C (Applications and Reviews). 2010; 40(1):

1–12.
[14] Karayiannis NB, Varughese B, Tao G, et al. Quantifying motion in video
recordings of neonatal seizures by regularized optical flow methods. IEEE
Transactions on Image Processing. 2005; 14(7): 890–903.
[15] Karayiannis NB, Sami A, Frost JD, et al. Automated extraction of temporal
motor activity signals from video recordings of neonatal seizures based on
adaptive block matching. IEEE Transactions on Biomedical Engineering.
2005; 52(4): 676–686.
[16] Ntonfo GMK, Ferrari G, Raheli R, et al. Low-complexity image processing
for real-time detection of neonatal clonic seizures. IEEE Transactions on
Information Technology in Biomedicine. 2012; 16(3): 375–382.
[17] Cattani L, Kouamou Ntonfo GM, Lofino F, et al. Maximum-likelihood
detection of neonatal clonic seizures by video image processing. In: 2014 8th
International Symposium on Medical Information and Communication
Technology (ISMICT); 2014. p. 1–5.
[18] Cattani L, Alinovi D, Ferrari G, et al. A wire-free, non-invasive, low-cost
video processing-based approach to neonatal apnoea detection. In: 2014
IEEE Workshop on Biometric Measurements and Systems for Security and
Medical Applications (BIOMS) Proceedings; 2014. p. 67–73.
[19] Kay SM. Fundamentals of Statistical Signal Processing: Estimation Theory.
1st ed. Upper Saddle River, NJ, USA: Prentice Hall; 1993.
[20] Solomon C and Breckon T. Fundamentals of Digital Image Processing. 1st
ed. Croydon, UK: Wiley-Blackwell; 2011.
[21] Swets J. Measuring the accuracy of diagnostic systems. Science. 1988; 240
(4857): 1285–93.
[22] Wu HY, Rubinstein M, Shih E, et al. Eulerian video magnification for
revealing subtle changes in the world. ACM Transactions on Graphics.
2012; 31(4).
[23] Alinovi D, Cattani L, Ferrari G, et al. Spatio-temporal video processing for
respiratory rate estimation. In: 2015 IEEE International Symposium on
Medical Measurements and Applications (MeMeA) Proceedings; 2015.
p. 12–17.
[24] Alinovi D, Ferrari G, Pisani F, et al. Respiratory rate monitoring by max-
imum likelihood video processing. In: 2016 IEEE International Symposium
on Signal Processing and Information Technology (ISSPIT); 2016.
p. 172–177.
 Chapter 15
IoT sensor networks in healthcare
Rinki Sharma1

Over the past few years, there have been numerous advances in the health-care
industry. Incorporation of Internet of Things (IoT) in the health-care applications
has unlocked numerous possibilities in the way healthcare operates now. This
chapter concentrates on IoT sensor networks in healthcare. The suitability of sensor
networks in healthcare is presented. The applications in which IoT sensor networks
play a crucial role, and thus related sensors are presented. Key sensors, and their
applications in healthcare, are presented. The most common and popular use cases
of IoT sensor networks in healthcare are discussed. The wireless communication
technologies used in IoT sensor networks for healthcare are described along with
their characteristics. Even though health-care IoT (H-IoT) networks have been in
use for some time, and their incorporation in health-care services is only expected
to increase further over the years, their implementation and adoption is still a
challenge. This chapter discusses those challenges in detail. The potential con-
temporary technologies that help in overcoming these challenges are explained.

List of abbreviations

3D three dimensional
EHR electronic health record
GPS global positioning system
H-IoT health-care Internet of Things
IoT Internet of Things
IR infrared
IrDA Infrared Data Association
LAN local area network
MEMS microelectromechanical systems
NFC near-field communication
NFV network function virtualization

1
Department of Computer Science and Engineering, Ramaiah University of Applied Sciences,
Bengaluru, India
284 The Internet of medical things

QoS quality of service

RFID radio frequency identification
SDN software-defined networking
UWB ultra-wideband
WBAN wireless body area network
Wi-Fi wireless fidelity
WSN wireless sensor networks

15.1 Introduction
In the past decades, the interaction of doctors with their patients remained limited to
the patients physically visiting the doctors/nurses and vice versa. With advances in
communication technologies and ease of communication, this interaction was
advanced to the form of tele- and text communications. The health-care technology
evolved further with the introduction of electronic health records (EHRs) that made it
possible for the hospitals to store and access the patient-centric information. The portal
technology allowed patients to log in to the health-care service provider’s website
taking a more active role by keeping a check on their health, accessing their medical
records and tracking their appointments. The cloud computing and big data technol-
ogies are driving the innovation into this field. However, the solution to attain constant
remote monitoring of the patient’s health and vital parameters was made possible
through the Internet of Things (IoT) technology that is powered by the wireless sensor
networks (WSN). Sensors convert physical parameters into signals that can be pro-
cessed further by other devices. When used in the form of wearable devices, these
sensors can be used to check and monitor the physiological parameters of the patient.
The values of these parameters include body temperature, heartbeat, pulse rate, blood
pressure, oxygen saturation and blood sugar. With the help of these sensor-based IoT
devices, real-time patient health monitoring and clinical feedback have become a
possibility. The rising adoption of wearable devices, emergence of connected care
solutions and implementation of digital technologies for healthcare are the main fac-
tors causing the surge in the demand for digital health-care solution and the expansion
of the digital health-care market. The IoT-based health-care market is segmented into
medical devices, system and software and medical services. The medical devices can
be wearable external devices or implanted medical devices, as well as stationary or
mobile medical devices used in the hospitals or health-care centers.
The WSN and IoT-based devices are majorly used for remote patient mon-
itoring, telemedicine, telesurgery, medical imaging, health-care apparatus and
medication management, smart health-care equipment, connected devices and
clinical operations. According to the survey carried out by Global Market Insights,
the global digital health market is increasing exponentially and is predicted to be a
500 billion dollar industry by the year 2025 [1]. According to the report published
by the “Data-Driven Investor” [2], the estimates show that the inclusion of WSN
and IoT-based devices in the health-care industry has saved the health-care industry
 IoT sensor networks in healthcare 285

up to 300-billion dollars, as the patients can now be monitored remotely decreasing

the hospital admission costs, need for space, beds and other necessary equipment
and staff to serve the visiting patients. According to the same report, 70% of the
best-selling wearables are meant for health and fitness tracking of the present IoT
device market share, and 40% of the IoT devices and applications are being used
for health-care applications in the year 2020. Companies such as Microsoft, Apple,
Phillips, General Electrical (GE), Wipro and many startups have invested in IoT-
based health-care research and development. Microsoft Azure’s cloud platform
facilitates numerous health-care applications and services. Smart health monitoring
and tracking devices and solutions developed by Apple, Philips and Wipro such as
smart watches, smart beds/mattresses ARE used for monitoring of vital parameters,
diagnostic imaging and medical implants.
This chapter presents the impact of IoT sensor network on present-day health-
care systems and solutions. The advantages and issues of incorporating IoT tech-
nology in healthcare are discussed. The sensors used for health-care applications,
use cases, supporting technologies, popular architectures, solutions/products,
research opportunities, challenges (security in particular) are also discussed.

15.2 Wireless sensor networks (WSN) and Internet of

Things (IoT)
Over the past few decades, the embedded computing platforms have evolved and
emerged into integrated processing, storage and communication units that support
wireless communication. The need for low-cost, low-powered and small-sized
computing systems in the medical field has witnessed the surge of WSN-based
applications and equipment in healthcare. The benefits of using sensor-based IoT
networks for health-care applications are presented in Figure 15.1.
These sensor-based devices can be worn, implanted or fixed on the stationary or
mobile medical devices. These sensor-based devices/systems have been proven
highly efficient in providing smart, reliable and real-time care by monitoring of
patient’s health and sharing their health records with the medical professionals. These

Low-power consumption

Significantly lower costs

Benefits of
sensor-based IoT Supports invasive applications
networks in
health-care Silicon is adaptable to body tissues
applications
Integration allows building of a range of systems

More number of systems can be incorporated

in single chip

Figure 15.1 Benefits of sensor-based IoT networks in health-care applications

286 The Internet of medical things

devices can be embedded in a variety of medical equipment not only at hospitals and
clinics but also at homes. Sensors meant for medical applications are used to monitor
physical and physiological parameters of the patient for detection, diagnosis, treat-
ment and posttreatment monitoring and management of patient’s health [3,4]. Some
of the popularly used sensors for medical applications are presented in Table 15.1.
Apart from the sensors, encoders are used in magnetic resonance imaging
machines, surgical robots, medical imaging, tomography, X-ray machines and
other critical/noncritical medical devices.
While these sensing devices have become popular for use in medical equip-
ment, they are constrained in resources such as computation and processing power,
battery power, storage/memory and supporting data rate/bandwidth [5]. The dis-
tinct features that have led to the rise of wireless sensing and communication
technology in healthcare are as follows:
1. Microelectromechanical systems (MEMS):
Advances in the MEMS technology have made it possible to implement and
deliver inexpensive remote patient monitoring and diagnostic capabilities. This
has made the early detection and curing of critical medical conditions. This has

Table 15.1 Key sensors and their applications in healthcare

Key sensors Applications

Pressure sensors Oxygen concentrators, anesthesia delivery machines,
ventilators, insulin pumps, blood-flow analyzers,
respiratory and blood pressure monitoring equipment,
surgical fluid maintenance, infusion pumps and
pressure-based dental instrument
Temperature sensors Medical incubators, neonatal ICUs, humidified oxygen
heaters, patient temperature monitoring units and
digital thermometers
Flow sensors Anesthesia delivery machines, ventilators, gas mixing
and electrosurgery wherein high-frequency electric
current is passed to destroy the malignant tissues
Image sensors Endoscopy, radiography, fluoroscopy, dental imaging,
mammography, retinal therapy and ocular surgery
Accelerometers Heart pacemakers, blood pressure monitors, defibrillators
and other integrated health-monitoring equipment
Biosensors Testing of cholesterol, blood glucose levels, pregnancy,
drug abuse and infectious diseases
SQUIDs (semiconductor quan- Comprise sensitive magnetometers and are used in
tum interface devices) magnetoencephalography (MEG) and magnetocardio-
graphy (MCG) systems used for analyzing neural
activity inside the brain
Implantable medical sensors Brain implants for treatment of neuropsychiatric dis-
orders, electrotherapeutic treatments, immunomodula-
tory implantable, bioresorbable medical devices for
repairing and regeneration of damaged cells and organs
and bioactive and drug-releasing surgical implants
 IoT sensor networks in healthcare 287

also made personalized care of patients by constantly monitoring their health

conditions and providing personalized treatment.
2. Small size and low cost:
The recent advances in MEMS, computing and processing technologies
have made real-time and reliable exchange of patients’ vital parameters and
medical records possible. The wearable, portable and ambulatory medical
sensors have made constant/frequent measurement of physiological parameters
possible without visiting the hospitals and health-care centers. The devices,
mainly wearable ones, are targeted toward fitness enthusiasts.
Implantable medical sensors are required to be small in size and are equipped
with radio frequency identification (RFID) or near-field communication (NFC)
communication modules to enable wireless communication and charging.
3. Connectivity and communication:
With the advances in communication technology, protocols, standards and
antenna design, the medical sensor motes are capable of communicating wireless
to other nearby devices or cloud. For short-range communication, these devices
are equipped with modules supporting technologies such as Bluetooth, Zigbee,
Infrared (IR) RFID and NFC. For long-distance communication, wireless fidelity
(Wi-Fi), cellular and cloud-computing services are used.
While supporting wireless, mobile and short/long-distance communication,
these medical sensors are also capable of recording the parameters/signals in non-
volatile memory and share it later when wireless connection is available. These
devices are also capable of communicating and uploading the recorded information
over cloud that can be accessed by the authorized personal over time while main-
taining patient history.
Existing medical sensor devices make use of 8- or 16-bit microcontrollers
having tens of kilobytes of random-access memory and hundreds of kilobytes of
read-only memory. These devices have small duty cycles and go to sleep mode
when not operational. They are equipped with low-power radios capable of trans-
mitting at the rates of up to 250 kbps and consume 20–60 mW of power for short-
distance (up to 10 m) communication.

15.3 Role of Internet of Things (IoT) sensor networks

in healthcare
Over the past decade, machine-to-machine and device-to-device communication
technologies have given networking and connectivity a different meaning by
enabling and establishing communication between devices. As the name suggests,
with IoT, things/devices are able to communicate with little or no human inter-
vention. With the level of automation achieved through IoT, this technology has
found numerous applications in the wide range of technologies such as healthcare,
emergency services, smart cities, vehicular networks, home and industrial auto-
mation, waste management and structural health monitoring. In this section, we are
going to concentrate on the applications of IoT in healthcare.
288 The Internet of medical things

The IoT sensor network-based health-care market is expected to rise with a

compound annual growth rate of 30.2% over the period of 2018–23. Some of the
key causes driving this industry are as follows:
1. advances in communication technologies and wireless communication in
particular,
2. wireless communication technologies with high data rates supporting reliable
and real-time communication,
3. advances in MEMS technology and biosensors related to health-care
applications,
4. availability of cloud/fog/edge devices for efficient processing and storage,
5. digitization of medical services and records such as EHRs,
6. advances in wearable health devices,
7. increasing geriatric population,
8. increasing need and demand for advanced health-care information systems.
Before the introduction of IoT sensor networks in healthcare, availability of an
efficient, reliable, responsive and remote health-care system was the challenge for
the health-care industry. With the rise in geriatric population and increasing pre-
valence of lifestyle and chronic diseases, the demand for advanced health-care
systems has increased. To cater to the demand of increasing population that needs
healthcare, the presently existing health-care system needs transformation.
Digitizing of the healthcare system, from consultations (patient–health-care pro-
fessional interaction) to maintenance of health records and management of health-
care equipment, stocks and the fleet (such as ambulance) [6] is the way to go for
handling these demands.
Figure 15.2 presents the health-care IoT (H-IoT) use cases. This section dis-
cusses the use cases of H-IoT. H-IoT is gaining popularity to provide remote
health-care services such as diagnosing, monitoring and remote surgeries over the
Internet [7]. Smart rehabilitation is used to take care of the aged. H-IoT connects
health-care resources and assistive devices with doctors, nurses, caretakers, hospi-
tals and rehabilitation centers. Some of the popular H-IoT applications as presented
in Figure 15.2 are as follows:

1. Remote patient monitoring: Wearable devices and sensors are used to monitor
the vital parameters of the patient. This information can be transmitted wire-
lessly to the doctor’s/nurse’s device or to the cloud from where it can be read by
the medical staff involved in taking care of the patient. Such a facility allows the
patient to move around instead of sitting at one place and allows doctors to
monitor patient’s vitals without requiring to visit the patient often [8].
2. Fitness and activity trackers: People wearing fitness and activity trackers
mounted on wrist, ankle or belt is a common sight these days. These trackers
check the physical activity of the user such as steps taken, stairs climbed and
vital health parameters such as pulse, heartbeat, body temperature and other
basic physiological parameters. This data is stored on the cloud and users can
study it over time to track their fitness and activity levels [9].
 IoT sensor networks in healthcare 289

Remote patient monitoring

Fitness and activity trackers

Remote medication
H-IoT use cases
Robotic surgery

Drug management

GPS smart sole

Patient care robot

Medical waste management

Body scanning

Figure 15.2 Health-care IoT (H-IoT) use cases

3. Remote medication through ingestible sensors: Ingestible sensors coated

with magnesium and copper are used to trigger a signal if medication is not
taken by a patient within appropriate time [10].
4. Robotic surgery: To perform surgical operations with enhanced control and
precision, surgeons use IoT-enabled robotic surgery devices. Surgeons can also
perform surgery in remote locations using these devices. However, this calls
for real-time response over a high bandwidth communication channel [11].
5. Drug management: Drug supply, distribution, management and monitoring
are carried out by using RFID. This helps in achieving enhanced supply chain
management and reduced overall production costs. In the case of wrong dis-
tribution, reports can be sent to the regulatory bodies through cloud [12].
6. Global positioning system (GPS) smart sole: A smart insole embedded with
GPS can be used to monitor the movement and location of the patient. This
application of H-IoT can be particularly useful to monitor the movement of
aged people or Alzheimer patients [13].
7. Body scanning: Smart body scanner is used for regular fitness checks as well
as to check the vitals of a person. A weight scale, also known as the turntable,
spins the body of the concerned person for 15 s. Three dimensional (3D)
cameras are used to capture 360-degree images of the person to develop a 3D
model of the person’s body. A body fat, circumference, vitals and weight scan
are sent on the mobile devices. A comparison of new readings with old values
of the parameters is also made available [14].
290 The Internet of medical things

8. Medical waste management: Medical waste disposal, monitoring and man-

agement are critical issues in healthcare. Automated medical waste manage-
ment uses sensors embedded in dustbins located at different locations in the
hospital to notify the quantity of waste. When the waste needs to be cleared, an
automated robot is used to collect the waste and dump it in an appropriate
place. This avoids any human interaction with hazardous medical waste and
keeps the environment clean [15].
9. Patient care robot: These robots are used to provide care and support to the
patients in the hospital. They are used to deliver medication, food and neces-
sary items to the patient. Equipped with onboard sensors, these robots navigate
using built-in maps and communicate over wireless medium [16].
Some of the key health-care applications based on IoT are tabulated in
Table 15.2.
With the incorporation of IoT in healthcare, some of the expected transfor-
mations are as follows:
1. Constant patient monitoring: Continuous monitoring of the vital parameters
of a patient even from remote locations is possible. The need for doctors/nurses
going on rounds or them meeting the patient in person is reduced. Need for
patients to take appointments from the doctors and waiting for doctor’s avail-
ability for consultation is not there anymore. Wireless body area networks
(WBAN) measure various vital parameters of the patient through concerned
sensors and transmit this information over the network to the smart devices
through which the concerned can monitor these parameters remotely and take
appropriate actions.
2. Auto patient access: Intelligent personal devices such as smartphones, wear-
able watches and personal digital assists equipped with healthcare-related
sensors facilitate the patient to keep a track of vital parameters and store the
information for future analyses.
3. Customized remote treatment: Customized treatment plans can be shared by
the doctors with the patients remotely, accessing the EHR of the patients and
information of the patient’s vitals over time.
4. Virtual care: Real-time information and notifications can be sent to the
patients regarding their medication, exercise and diet as part of virtual con-
sultation. This helps the health-care centers to advance treatment and achieve
improved outcomes from the treatment of their patients.
5. Remote retail centers: Medicines and other necessary medical equipment can
be ordered and paid for automatically without the need to visit the medical
store. Automation provided by IoT can be used to identify the medicines
required to be ordered based on need/requirement.
6. Connected care: The vitals measured by the biosensors are shared and stored
over cloud, the consultations provided by the health-care personal.
7. Health insurance: The cost of healthcare can be eased through easy and
reliable health insurance system that can be established for easy risk assess-
ment, pricing and claim handling, with transparency.
 IoT sensor networks in healthcare 291

Table 15.2 IoT-based healthcare

Medical condition Sensors and connections Operation

Body Wearable body temperature These devices transmit the
temperature sensors and WBAN-connected measured temperature to the
smart devices cloud through an appropriate
Heart rate Capacitive electrodes on a PCB gateway. This information can
monitoring connected to the wireless be accessed remotely through
transmitter in real time the cloud
Blood pressure Wearable blood-pressure-
monitoring monitoring sensors, automatic
inflation and measurement data
are collected, analyzed and
transmitted through connected
smart devices
Oxygen saturation Wearable pulse oximeter sensor
monitoring measures pulse rate and does
pulse-by-pulse oxygen satura-
tion measurement
ECG self- Device/application registers ECG
monitoring data to the inbuilt ECG self-
check software
Eye disorders Visual inspection and pattern- Cloud-aided applications and re-
matching devices comprising a lated software platforms used
standard library of images used for monitoring and diagnosis
to check eye disorders
Skin allergies/ Smartphone cameras, pattern
infection matching and remote visual
inspection
Pulmonary Calculation of air-flow rate,
infections flow-time, volume-time and
flow-volume graphs. Built-in
microphones in the smartphone
are used for this
Cough and bron- Analysis of spectrograms and their
chial infections classification through machine
learning algorithms. Built-in
microphones in the smartphone
are used for this
Allergic rhinitis Speech recognition and classifi-
cation using machine learning
techniques. Built-in micro-
phones in the smartphone are
used for this
Carcinoma melano- Smartphone camera used to cap-
ma detection ture images. Pattern matching
through a library of images of
cancerous skin
Wound analysis Smartphone camera used to cap-
ture images that are observed
remotely to analyze the wounds
and suggest corrective action
(Continues)
292 The Internet of medical things

Table 15.2 (Continued)

Medical condition Sensors and connections Operation

Rehabilitation WBAN-connected smart devices Interactive wireless network tech-
used for measuring patient’s nologies used through remote
vital parameters, patient track- locations to interact with the
ing, monitoring, event detec- patient and for consultation
tion, notification and reporting
Drug management Diagnosis and prognosis of the Accessing the vitals of the patient
vitals measured and transmitted and consultation and suggestion
through the WBAN sensors and for medication are done from
stored in the cloud through remote location
wireless communication
Medicines can be identified
through RFID and their quantity
monitored through sensors.
Alert/notification generated for
refill Alert/notification gener-
ated to inform the patient about
right time and dosage of the
medicine
Wheelchair naviga- Proximity sensors, location mon- Heterogeneous wireless commu-
tion itoring, LIDAR/RADAR, object nication and network technolo-
detection, avoidance and navi- gies used for navigation and
gation, lane detection and navi- monitoring
gation. Alert/notification/alarm
generated in unavoidable cir-
cumstances
Remote robotic High-definition video streaming, as- High data rate wireless commu-
surgery sistance robots, smart sensors and nication technologies such as
wearables and connected ambu- 5G, robotics and automation
lance communicate with parame- used to make this a reality
dics at the hospital premises

15.4 Communication technologies for health-care IoT

sensor networks
The healthcare-related IoT communications require reliable and, in many applica-
tions, real-time communication. To support such communication, there is a need
for tested and reliable technologies that are capable of supporting communication
with varying data rates, communication range, operating frequency, transmission
power, communication power and security. Based on the application requirement
and services provided by the communication technologies, appropriate technology
is used for H-IoT sensor networks. The health-care applications related to com-
munications vary from personal area network that is popularly used by sensor
network technologies, local area network (LAN), metropolitan area network to
wide-area-network-based wireless communication technologies. This section dis-
cusses the communication technologies used for H-IoT applications. These tech-
nologies and their characteristics are presented in Table 15.3.
Table 15.3 Communication technologies used for H-IoT sensor networks

Wi-Fi Zigbee Bluetooth RFID IrDA UWB

International Standard IEEE 802.11a/b/g/n IEEE 802.15.4 IEEE 802.15.1 ISO 18047 – IEEE 802.15.3a
Data rate 1–450 Mbps 20–250 kbps Up to 3 Mbps 106–424 kbps 14.4 kbps 53–480 Mbps
Frequency band of operation 2.4 GHz, 5 GHz 868 MHz, 2.4 GHz 2.4 GHz 13.56 MHz 850–900 nm 3.1–10.6 GHz
Communication distance 200 m 10–20 m 20–200 m 20 cm 0–1 m 0–10 m
Communication type Point to multipoint Point to multipoint Point to point Point to point Point to point Point to point
Security SSID, WEP, WPA AES AES, ECDH AES – AES
Power (mW) >1 000 mW 100 mW 1–100 mW <1 mW 1 mW <1 mW
294 The Internet of medical things

By using Wi-Fi technologies, the medical and IoT devices all over the campus
can be connected for reliable communication by virtue of their communication
range and supported data rate. Wi-Fi supports device mobility over the network,
thus providing a dynamic network environment.
Zigbee is a popular WSN technology that uses unlicensed wireless spectrum.
Various medical sensors make use of Zigbee for communication. In H-IoT, the
most common use of Zigbee is in remote patient monitoring. Body sensor network
is used for patient health monitoring, and fitness and activity trackers make the use
of Zigbee sensors. Zigbee provides an appropriate balance of parameters such as
data rate, communication range and transmission power to make it the chosen
technology for these applications.
Bluetooth technology supports both voice and data, provides 1 Mbps of data
rate and uses unlicensed spectrum for communication. These features of Bluetooth
make it suitable choice for H-IoT. In a hospital campus, Bluetooth devices can be
identified and visitors can be informed about the floor plan, doctors on duty and
appointment scheduling through Bluetooth beacons. Bluetooth can be used for
automated patient check-in and checkout, optimized patient flow, scheduling
appointments with doctors, compliance tracking and recording, asset tracking and
wayfinding.
RFID is a short-distance, noncontact communication technology, which does
not require a direct line of sight communication. It is an economical and reliable
technology used for location and identification of objects. In H-IoT applications,
RFID is extensively used to identify, locate and track medical equipment.
Infrared Data Association (IrDA) is a very short-distance, point-to-point,
line of sight, which involves very low bit rate wireless optical communication
technology. This is used to carry out physically secured data transmission. IR
technology is used in remote control of appliances such as television, projector, air
conditioner and such other devices. In healthcare, IrDA can be used for remotely
controlling different medical devices.
Ultra-wideband (UWB) is a short-distance communication technology that
provides data rates of up to 450 Mbps. This uses low-power pulses for commu-
nication and generates low electromagnetic radiation, thus making it suitable for
medical applications. UWB radar is used for the monitoring of patient’s motion
over short distances, real-time exchange of medical images (such as X-rays, car-
diology-, pneumology-, obstetrics-, ear, nose, throat-imaging) and data over a dis-
tance of up to 10 m.

15.5 Challenges in the implementation of H-IoT and

related research
With constant increase in the need for H-IoT applications and uses, the integration
of H-IoT in health-care organizations and other related areas/applications has risen.
However, though there are numerous benefits of integrating IoT sensor networks
with healthcare, it faces numerous challenges also, that need to be overcome. This
 IoT sensor networks in healthcare 295

section discusses the challenges involved in the integration of H-IoT sensor net-
work in healthcare.

1. Massive data generation: IoT sensor networks in healthcare involve thou-

sands of devices involved in various healthcare-related applications and gen-
erate huge amounts of data that mostly needs to be stored and processed in real
time. This calls for the development of a large-scale health-care IoT infra-
structure that requires financial investment, time and effort. Solutions provided
in [17,18] deal with handling big data generated by health-care applications.
2. Need for updated infrastructure: With continuous upgradation of IoT devices,
the IT infrastructure also needs to be upgraded constantly. Out-of-date infra-
structure and software become incompatible with new and upgraded devices.
This requires the health-care organizations to constantly upgrade their IT and
software infrastructure so that it is compatible with changing IoT-based health-
care technology. Authors in [19] have developed a software-based IoT health-
care infrastructure using technologies such as network virtualization, cloud and
blockchain.
3. Need for upgraded network technologies: The rise of communicating IoT
devices has led to exponential rise in the network traffic. To deal with this
exponential rise of data and maintain the quality of service (QoS) required by
real-time and reliable health-care applications, there is a need for high network
bandwidth, computation and storage. As these devices mostly communicate
over wireless networks, when the density of the IoT devices increases within a
given area, the network performance may drop due to node mobility and inter-
ference among neighboring devices. Traditional network architectures are not
considered to be appropriate for these IoT sensor-based mobile networks. To
support the high data rate and bandwidth requirements of these networks, dual
polarized directional antenna-based protocols presented in [20–22] can be used.
4. Heterogeneous networks: Implementation of IoT sensor networks for
healthcare involves a variety of networks and requires devices implemented
on varying hardware, operating systems and programming languages to
coexist and interoperate. The communication technologies used cover short
range such as Bluetooth, IrDA, UWB and Zigbee, as well as long range such
as Wi-Fi and global system for mobile communication/long-term evolution.
Long-range communication leads to the device battery draining out faster. It
is necessary that despite this heterogeneity in implementation and commu-
nication, these devices should be able to interoperate by identifying and dis-
covering each other. To deal with the issues arising due to IoT network
heterogeneity, authors in [23] have developed a framework to decompose the
IoT applications into smaller operations that can be handled independently.
Authors in [24] have developed semantic interoperability model to ensure
interoperability between heterogeneous IoT devices and technologies.
5. Quality of service (QoS): IoT sensor networks are characterized by low-
latency, low-power operation with high reliability. As H-IoT deals with cri-
tical issues involving a patient’s health, they are bound to be time-critical by
296 The Internet of medical things

nature. Therefore, minimal delays should be incurred in transmission and

processing of H-IoT signals. This can be achieved by the availability of high
bandwidth networking resources. Reliable network connectivity is essential in
IoT sensor network for health-care applications that require real-time inter-
action and data access, such as remote surgery. To maintain QoS of health-
care sensor network, it is required to maintain sufficient bandwidth and data
rate. The solutions are presented in [20–22]. An adaptive QoS aware algo-
rithm is presented in [25] for QoS computation of different IoT-based health-
care applications.
6. Scalability: The growing popularity of H-IoT has led to exponential increase
in the number of H-IoT devices. It is important to maintain consistent per-
formance and QoS in the H-IoT network even as the new devices are added to
the network. To support a large-scale deployment of H-IoT networks, this
issue needs to be addressed effectively to avoid system downtime particularly
in crucial H-IoT networks. A fault-tolerant and scalable architecture for IoT-
based health-care networks is presented in [26] that is capable of handling
situations such as node malfunction and traffic bottleneck due to high node
density and data.
7. Data collection and management: IoT sensor devices generating digital
health-care data face numerous data collection and management challenges.
As the state of a human body changes constantly, thousands of bytes of data
are generated every minute. A connected health-care network comprises
heterogenous network technologies, devices and systems, leading to large
data volumes, variety, velocity and veracity. Managing and analyzing this
data requires data-driven learning techniques. Also, the data collection and
management techniques need to be standardized for health-care applications.
For health-care applications, data integrity also plays an important role;
therefore, robust authentication systems are required to be in place for health-
care systems. Authors in [27] have presented an architecture for integration of
WSN with cloud service for efficient collection of IoT sensor network data
and sharing.
8. Security and privacy: Security and privacy of the patients and their data is of
paramount importance in H-IoT. In many cases, H-IoT data is communicated
for long distances over heterogeneous networks and stored in the cloud,
making it vulnerable to attacks. Therefore, robust encryption and authenti-
cation techniques for security and privacy of data are essential. However,
while designing these techniques, it has to be taken into consideration that the
devices used for H-IoT are resource constrained; hence, the developed solu-
tions too should be lightweight and energy efficient. A survey of security and
privacy issues in H-IoT is presented in [28]. Advanced technologies such as
learning-based Deep Q-Networks are used to develop solutions to attain
security and privacy in health-care networks [29].
9. Increased attack surfaces: IoT sensor networks give rise to numerous vul-
nerable security areas. This is risky for medical/health-care applications, as
hackers can hack the medical devices, log in to these devices and steal/modify
 IoT sensor networks in healthcare 297

the information, which can lead to life critical situations. An entire hospital
network can be hacked through the infamous ransomware attack. Authors in
[30] have developed a framework to test the security of IoT devices. An
intrusion detection system for H-IoT networks is presented in [31].
10. Cost: While going digital and using technology in healthcare has brought down
the cost of physical visits for both patients and health-care professionals, as
well as providing ease of interaction/consultation, and patient health observa-
tion, the IoT sensor network technology is still out of reach for many. While
IoT-based healthcare is fascinating and promising, its implementation is yet to
overcome the cost considerations. The successful development, implementation
and optimization of these technologies are still an issue and require setting up a
costly infrastructure to ensure reliable connectivity and communication. Some
of the low-cost smart health-care solutions are presented in [32,33].

15.6 Contemporary technologies to overcome the

challenges on IoT sensor networks for healthcare

With the evolution of IoT and other mobile and wireless communication technol-
ogies, traditional computer networking technologies also need to be evolved, in
order to cater to the demand of contemporary applications such as IoT sensor
networks for healthcare. This section presents some of those contemporary tech-
nologies being used extensively in IoT sensor networks for healthcare.

15.6.1 Cloud/fog/edge computing for H-IoT sensor networks

The H-IoT devices produce massive amounts of data by a minute. This data needs to
be processed efficiently and reliably. In applications such as robotic surgery,
wheelchair navigation, constant patient monitoring and live consultations, the data
needs to be processed in real time with high reliability, high efficiency and least
delay. Over the years, cloud computing has been the go to technology for IoT devices
to store and process the data. In such cases, the IoT data is transmitted all the way to
the cloud and processing and storage. When stored in the cloud, the data can also be
accessed/downloaded for future reference. From the IoT perspective, the massive
data generated by these devices is stored and computed securely and efficiently over
the cloud and is accessible seamlessly over multiple devices [34]. Amalgamation of
cloud with network function virtualization (NFV) further reduces the capital and
operation expenses of the H-IoT sensor networks [35,36]. The data transmitted to the
cloud and stored over it is end-to-end encrypted. However, cloud computing for H-
IoT requires availability and access to the massive data centers located in remote
locations. These data centers can become inaccessible due to unavailability of proper
Internet connectivity and cause delay in communication and processing of the IoT
data due to distance between the IoT devices’ cloud/data centers. Such delays can
lead to poor quality of experience and slow data processing, thus adversely affecting
the performance of real-time H-IoT applications, which is unacceptable and critical
for life health-care applications in particular.
298 The Internet of medical things

For efficient, reliable and real-time processing of delay-sensitive H-IoT data,

edge and fog computing technologies are used. Both edge and fog computing are
the extensions of cloud computing. While cloud computing allows the health-care
organizations such as hospitals, nursing homes and rehabilitation centers to not
spend on storage and processing infrastructure, it is observed over the years a
combination of edge, fog and cloud computing benefits through efficient usage of
end-user IoT devices and faster computing and processing of the health-care data.
Better distribution of data for storage and processing achieves better performance
by reducing the network traffic to the cloud, thus enhancing its operational effi-
ciency. The edge and fog computing networks enhance network performance and
optimize usage of cloud-computing resources by performing computations at the
edge/fog nodes themselves, reducing network traffic and thus reducing the risk of
network connection bottleneck [34,37,38]. It also increases network security by
encrypting the data closer to the network devices.
Both edge and fog computing bring the location of computation and processing
closer to the end nodes, thus drastically reducing the delays involved in delay-
sensitive health-care applications. The edge computing places processing and
computation power on the edge devices such as embedded controllers on the
health-care devices, while fog computing places the computation and intelligence
power on the network gateways and devices in the LAN.
Edge computing in H-IoT is powered by the fact that there are numerous edge
devices in an IoT network, generating a massive amount of data in the network that
leads to challenges in processing this data over a data center located remotely. In
edge computing, the data is sent to the edge device, and it is processed on the edge
device itself. Edge computing comprises thousands of nodes, wherein each device
is capable of acting as the server in edge network. Edge network is a consolidation
of voluntary compute devices that are located one to two hops away from the sensor
nodes. These devices support high mobility but have low compute power.
Therefore, edge computing is the least vulnerable form of decentralized processing
and storage. Some of the advantages of edge computing are as follows:
1. faster data processing
2. reduced network traffic
3. real-time data analysis
4. reduced network operating cost
5. decentralized data processing and storage
6. reduced network vulnerability
While there are advantages of edge computing, it is not possible to run all the
applications in intelligent endpoints as the endpoints face challenges regarding
security, reliability, modularity, storage, processing power and environment.
Therefore, there is a need for fog and cloud computing (Figure 15.3).
Fog computing is useful in the case of intermittent network connectivity. In
fog computing, the delay-sensitive requests are transmitted directly to the fog
computing devices and are processed within the LAN. The data from applications
that are not/less delay sensitive is sent to the cloud data centers for processing. The
 IoT sensor networks in healthcare 299

• Data warehousing
• Remote data access
Cloud data centers • Big data processing and
(in thousands) analytics
• Al and ML
• Centralized services

• Computational nodes
• Gateway nodes
• On local networks
Fog nodes • Data analysis and
(in millions) reduction
• Virtualization
• Correlation and control

• IoT devices / sensors

(end devices)
Edge devices • Real-time data processing
(in millions) • On-premises data
visualization
• Micro-data storage

Figure 15.3 Edge, fog and cloud for H-IoT sensor networks

devices in the fog network have high compute power compared to the edge network
devices. These are installed as a backbone to the cloud networks. These devices are
located five to six hops away from the sensor nodes and support high-speed con-
nectivity and guaranteed connectivity to the edge devices. Some of the advantages
of fog computing are as follows:
1. faster data computing and processing
2. reduced latency
3. enhanced data security
4. setter network and data control
5. support for online and offline data access
6. decentralized data storage and processing
By adding layers of edge and fog nodes, the network load is partitioned to run
at the optimal network level.
Cloud computing provides enhanced scalability and performance to the IoT
networks. As the end H-IoT devices generate huge amounts of data every second, it is
difficult for the enterprises and health-care organizations to store such huge data for
future access. Due to the issues such as node mobility, node’s geographic location,
latency, network bandwidth and reliability, it is neither efficient nor advisable to run
all the applications in the cloud. That is where edge and fog computing play an
important role. Cloud comprises data centers that are capable of data warehousing, big
data processing and making the data available for future remote access. Cloud pro-
vides Infrastructure as a Service for better H-IoT scalability, storage and processing
power. Some of the benefits of cloud computing for H-IoT networks are as follows:
1. data storage,
2. access to data remotely for future use,
3. remote data processing,
300 The Internet of medical things

4. data security,
5. reduced infrastructure cost at the organization,
6. present-day cloud solutions capable of applying artificial intelligence, machine
learning and big data techniques on the IoT data.

15.6.2 Software-defined networking for H-IoT sensor

networks
Software-defined networking (SDN) separates the control plane (the decision-
making plane, mostly implemented in software) from the data plane (the data
transfer plane, mostly implemented in hardware). The application plane comprises
the applications configured by the network operator such as load balancer, access
control and bandwidth allocation [39]. The SDN network controller has the global
view of the network and is capable of making decisions about resource allocation
based on changing network requirements. The traffic forwarding policies are
defined by the network operator, while the centralized network controller facilitates
automated network management. To handle the increasing network demand, net-
work resources can be virtualized. NFV in association with SDN brings in
numerous advantages to network performance while achieving reduced capital
expenditure and operational expenditure [36]. The communication between net-
work devices and applications takes place through the network controller. As the
networks scale up, the number of network controllers incorporated into the network
could be increased to maintain the network performance with increasing
traffic load.
An example of SDN-based H-IoT [40] system is presented in Figure 15.4. The
system is divided into three following planes:
Data plane: The data plane comprises the network devices and infrastructure
used to forward the data over the network. This plane is also called the forwarding
plane. The network routers and switches fall in this plane. As shown in Figure 15.4,

Data plane
Data generating IoT devices and
network devices such as switches

Control plane
controler and cloud

Application plane
IoT sensor network healthcare
applications

Figure 15.4 SDN for H-IoT sensor networks

 IoT sensor networks in healthcare 301

the data is acquired from sensors and other IoT devices through the wireless/wired
network infrastructure. The received data is routed and forwarded to the network
controller (in the control plane) through the data plane.
Control plane: Control plane is the decision-making plane. The main entity of
this plane is the network controller that comprises the logic written for traffic control
such as routing, load balancing and traffic engineering. The controller fetches,
maintains different network information, such as network state, topology and statis-
tics, and makes network infrastructure control decisions based on this information.
The network controller receives application requirements from the application layer
and makes decision regarding QoS and dynamic bandwidth allocation. It has an
abstract view of the network, the events in the network and network statistics.
The SDN can be combined with the cloud for better network management and
network data access. This removes the barrier associated with network hardware
access and builds the network implementation cost-effective and agile. As seen
from Figure 15.4, the data from the network controller is uploaded onto the cloud.
Similarly, the network controller also fetches the data from the cloud and transmits
it to the end devices.
Application plane: Application plane presented in Figure 15.4 comprises all
the healthcare-related applications that work with the H-IoT data. The application
plane comprises the receiving applications that obtain the data originated at the
patient. This data is used for diagnosis, analytics, generate statistics and makes
appropriate decisions. As seen from Figure 15.4, the decisions made or messages
can be conveyed back to the patient through the cloud and network controller.

15.7 Conclusion
This chapter presents the role, significance, applications, supporting technologies,
challenges faced and contemporary technologies to overcome those challenges in IoT
sensor network for healthcare. IoT sensor networks play a significant role in
healthcare and have become popular in the health-care industry over the past few
years. According to different surveys and estimates, the use of IoT sensor networks in
healthcare will grow exponentially in the future. This chapter discusses the sensors
used in different healthcare-related applications. The communication technologies
facilitating the H-IoT networks and their characteristics are discussed. Despite
growing popularity of IoT sensor networks in healthcare, there are numerous chal-
lenges in the implementation and realization of H-IoT. These challenges are dis-
cussed in detail and the role of contemporary technologies such as edge/fog/cloud
computing and SDN to overcome some of these challenges is highlighted.

References
[1] https://www.businessinsider.in/science/news/iot-healthcare-in-2020-compa-
nies-devices-use-cases-and-market-stats/articleshow/74126142.cms: acces-
sed on 7 January 2021.
302 The Internet of medical things

[2] https://medium.com/datadriveninvestor/7-staggering-stats-on-healthcare-iot-
innovation-fe6b92774a5c: accessed on 7 January 2021.
[3] Sharma R., Gupta S.K., Suhas K.K. and Kashyap G.S., 2014, April.
Performance analysis of Zigbee based wireless sensor network for remote
patient monitoring. In 2014 Fourth International Conference on
Communication Systems and Network Technologies (pp. 58–62). IEEE.
[4] Al-Khafajiy M., Baker T., Chalmers C., et al., 2019. Remote health mon-
itoring of elderly through wearable sensors. Multimedia Tools and
Applications, 78(17), pp. 24681–24706.
[5] Sharama R., Shankar J.U. and Rajan S.T., 2014. Effect of number of active
nodes and inter-node distance on the performance of wireless sensor net-
works. In 2014 Fourth International Conference on Communication Systems
and Network Technologies (pp. 69–73). IEEE.
[6] Juneja S., Juneja A. and Anand R., 2020. Healthcare 4.0-digitizing health-
care using big data for performance improvisation. Journal of
Computational and Theoretical Nanoscience, 17(9 10), pp. 4408–4410.
[7] Tarouco L.M.R., Bertholdo L.M., Granville L.Z., et al., 2012. Internet of
Things in healthcare: interoperability and security issues. In 2012 IEEE
International Conference on Communications (ICC) (pp. 6121–6125). IEEE.
[8] Marakhimov A. and Joo J., 2017. Consumer adaptation and infusion of
wearable devices for healthcare. Computers in Human Behavior, 76,
pp. 135–148.
[9] Philip V., Suman V.K., Menon V.G. and Dhanya K.A., 2017. A review on
latest internet of things based healthcare applications. International Journal
of Computer Science and Information Security, 15(1), p. 248.
[10] Lomotey R.K., Pry J. and Sriramoju S., 2017. Wearable IoT data stream
traceability in a distributed health information system. Pervasive and Mobile
Computing, 40, pp. 692–707.
[11] Shabana N. and Velmathi G., 2018. Advanced tele-surgery with IoT
approach. In Intelligent Embedded Systems (pp. 17–24). Springer,
Singapore.
[12] Liu L. and Jia W., 2010, September. Business model for drug supply chain
based on the internet of things. In 2010 2nd IEEE International Conference
on Network Infrastructure and Digital Content (pp. 982–986). IEEE.
[13] Wilden J., Chandrakar A., Ashok A. and Prasad N., 2017. IoT based wear-
able smart insole. In 2017 Global Wireless Summit (GWS) (pp. 186–192).
IEEE.
[14] Muneer A., Fati S.M. and Fuddah S., 2020. Smart health monitoring system
using IoT based smart fitness mirror. Telkomnika, 18(1), pp. 317–331.
[15] Raundale P., Gadagi S. and Acharya C., 2017. IoT based biomedical waste
classification, quantification and management. In 2017 International
Conference on Computing Methodologies and Communication (ICCMC)
(pp. 487–490). IEEE.
[16] Patel A.R., Patel R.S., Singh N.M. and Kazi F.S., 2017. Vitality of robotics
in healthcare industry: an Internet of Things (IoT) perspective. In Internet of
 IoT sensor networks in healthcare 303

Things and Big Data Technologies for Next Generation Healthcare (pp.
91–109). Springer, Cham.
[17] Chen M.Y., Lughofer E.D. and Polikar R., 2018. Big data and situation-
aware technology for smarter healthcare. Journal of Medical and Biological
Engineering, 38, pp. 845–846.
[18] Manogaran G., Lopez D., Thota C., Abbas K.M., Pyne S. and Sundarasekar
R., 2017. Big data analytics in healthcare Internet of Things. In Innovative
Healthcare Systems for the 21st Century (pp. 263–284). Springer, Cham.
[19] Salahuddin M.A., Al-Fuqaha A., Guizani M., Shuaib K. and Sallabi F., 2017.
Softwarization of internet of things infrastructure for secure and smart
Healthcare. Computer, 50(7), pp. 74–79. doi:10.1109/MC.2017.195.
[20] Rinki S., 2014. Simulation studies on effects of dual polarisation and
directivity of antennas on the performance of MANETs (Doctoral disserta-
tion, Ph.D. thesis, Coventry University, UK).
[21] Sharma R., Kadambi G.R., Vershinin Y.A. and Mukundan K.N., 2015.
Multipath routing protocol to support dual polarised directional communication
for performance enhancement of MANETs. In 2015 Fifth International
Conference on Communication Systems and Network Technologies (pp.
258–262). IEEE.
[22] Sharma R., Kadambi G.R., Vershinin Y.A. and Mukundan K.N., 2015. Dual
polarised directional communication based medium access control protocol
for performance enhancement of MANETs. In 2015 Fifth International
Conference on Communication Systems and Network Technologies
(pp. 185–189). IEEE.
[23] Jha D.N., Michalák P., Wen Z., Ranjan R. and Watson P., 2019.
Multiobjective deployment of data analysis operations in heterogeneous IoT
infrastructure. IEEE Transactions on Industrial Informatics, 16(11),
pp. 7014–7024.
[24] Ullah F., Habib M.A., Farhan M., Khalid S., Durrani M.Y. and Jabbar S.,
2017. Semantic interoperability for big-data in heterogeneous IoT infra-
structure for healthcare. Sustainable Cities and Society, 34, pp. 90–96.
[25] Sodhro A.H., Malokani A.S., Sodhro G.H., Muzammal M. and Zongwei
L., 2020. An adaptive QoS computation for medical data processing in
intelligent healthcare applications. Neural Computing and Applications,
32(3), pp. 723–734.
[26] Gia T.N., Rahmani A.M., Westerlund T., Liljeberg P. and Tenhunen H.,
2015. Fault tolerant and scalable IoT-based architecture for health monitor-
ing. In 2015 IEEE Sensors Applications Symposium (SAS) (pp. 1–6). IEEE.
[27] Piyare R. and Lee S.R., 2013. Towards Internet of Things (IoTs): integration
of wireless sensor network to cloud services for data collection and sharing.
International Journal of Computer Networks & Communications, 5(5),
pp. 59–72.
[28] Hathaliya J.J. and Tanwar S., 2020. An exhaustive survey on security and
privacy issues in Healthcare 4.0. Computer Communications, 153, pp. 311–335.
304 The Internet of medical things

[29] Shakeel P.M., Baskar S., Dhulipala V.S., Mishra S. and Jaber M.M., 2018.
Maintaining security and privacy in health care system using learning based
deep-Q-networks. Journal of Medical Systems, 42(10), p. 186.
[30] Lally G. and Sgandurra D., 2018. Towards a framework for testing the
security of IoT devices consistently. In International Workshop on Emerging
Technologies for Authorization and Authentication (pp. 88–102). Springer,
Cham.
[31] Thamilarasu G., Odesile A. and Hoang A., 2020. An intrusion detection
system for internet of medical things. IEEE Access, 8, pp. 181560–181576.
[32] Gia T.N., Jiang M., Sarker V.K., et al., 2017. Low-cost fog-assisted health-
care IoT system with energy-efficient sensor nodes. In 2017 13th
International Wireless Communications and Mobile Computing Conference
(IWCMC) (pp. 1765–1770). IEEE.
[33] Ganesh G.R.D., Jaidurgamohan K., Srinu V., Kancharla C.R. and Suresh S.
V., 2016, December. Design of a low cost smart chair for telemedicine and
IoT based health monitoring: an open source technology to facilitate better
healthcare. In 2016 11th International Conference on Industrial and
Information Systems (ICIIS) (pp. 89–94). IEEE.
[34] Dang L.M., Piran M., Han D., Min K. and Moon H., 2019. A survey on
internet of things and cloud computing for healthcare. Electronics, 8(7),
p. 768.
[35] Ananth M.D. and Sharma R., 2017. Cost and performance analysis of net-
work function virtualization based cloud systems. In 2017 IEEE 7th
International Advance Computing Conference (IACC) (pp. 70–74). IEEE.
[36] Ananth M.D. and Sharma R., 2016. Cloud management using network function
virtualization to reduce CAPEX and OPEX. In 2016 8th International
Conference on Computational Intelligence and Communication Networks
(CICN) (pp. 43–47). IEEE.
[37] Ray P.P., Dash D. and De D., 2019. Edge computing for Internet of Things: a
survey, e-healthcare case study and future direction. Journal of Network and
Computer Applications, 140, pp. 1–22.
[38] Paul A., Pinjari H., Hong W.H., Seo H.C. and Rho S., 2018. Fog computing-
based IoT for health monitoring system. Journal of Sensors, 2018.
[39] Sharma R. and Reddy H., 2019. Effect of load balancer on software-defined
networking (SDN) based Cloud. In 2019 IEEE 16th India Council
International Conference (INDICON) (pp. 1–4). IEEE.
[40] Sharma R. 2021. 11 Software-defined networking for healthcare Internet of
Things. In Applications of Machine Learning in Big-Data Analytics and
Cloud Computing (pp. 231–247). River Publishers.
 Chapter 16
Machine learning for Healthcare 4.0:
technologies, algorithms, vulnerabilities, and
proposed solutions
Saumya1 and Bharat Bhushan1

Healthcare 4.0 is motivated from Industrie 4.0. It is a boon to the present health-
care system due to its widespread applications that have boosted its efficiency and
enhanced its services. Healthcare 4.0 is a vision that integrates all the leading
technologies together given that each technology has different benefits to offer to
the system. Various technologies such as big data, Health Cloud (HC), Health Fog
(HF), Internet of Things (IoT), blockchain, and machine learning (ML) are incor-
porated in Healthcare 4.0. There are many applications of Healthcare 4.0 for
patients, health-care professionals, resource management, etc. This chapter aims to
study ML with respect to Healthcare 4.0 and highlights different ML algorithms
and its applications for healthcare in different phases such as prognosis, diagnosis,
treatment, and clinical workflow. ML provides many solutions for Healthcare 4.0
which have applicability for patients, health-care professionals, as well as health-
care facilities. However, there is some vulnerability for ML in healthcare that needs
to be checked. This chapter highlights these vulnerabilities and presents the
recently proposed solutions in this regard.

16.1 Introduction

Healthcare today is covered with many challenges and difficulties such as the high
cost of health-care services, lack of skilled professionals, demands of high-quality
services, lack of collaboration among different health-care service providers, and
high competition among them [1]. These things are raising questions on the relia-
bility of the system for the patients and their satisfaction. Therefore, there is a need to
look for new solutions to overcome these challenges [2]. The need of the hour is to
evolve the health-care system, one such sector evolving is Industry Sector 4.0 which
is developed from Industries 1.0, 2.0, and 3.0. The goal of Industry 4.0 is to build on
automated smart machines that make the system cost-effective, enhance operation,

1
School of Engineering and Technology (SET), Sharda University, Greater Noida, India
306 The Internet of medical things

and increase quality of services. These principles taken from Industry 4.0 are applied
to the health-care system and gave birth to Healthcare 4.0, without any existence of
Healthcare 1.0, 2.0, and 3.0. It aims to enhance the efficiency of health-care services,
including new technologies to increase the quality of services and still being cost-
effective for the patients. The benefits of Healthcare 4.0 are numerous such as
improve flexibility, reliability, scalability, and cost-effectiveness [3]. Apart from this,
it will also help in better management and response to health pandemics like COVID-
19 [4]. Despite its benefits, building such a versatile system for healthcare is chal-
lenging and complex within itself. Many factors are needed to be considered. But,
once built, it will boost its efficacy. Healthcare 4.0 is built on the integration of
human resources, expertise with technologies like blockchain, ML, HC, big data,
IoT, deep learning (DL), and artificial intelligence (AI).
The chapter discusses ML, its algorithms, and its applicability to Healthcare
4.0. ML is a vast field, which is growing every day. It has already transformed
many systems such as governance, transportation, and manufacturing. The use of
these technologies is carried in people’s day-to-day life either knowingly or
unknowingly. ML is now being incorporated into the field of healthcare, which is
influencing this system to provide various advantages [5]. Using ML models in
healthcare has many benefits like it will make the system less dependent on human
resources, fewer chances of errors and thus, increase in efficiency without manual
support. It can assist medical professionals through AI in learning information from
medical books, texts, etc. Medical tasks like monitoring, analyzing, and managing
reports of patient become efficient when embedding of ML with IoT devices is
done [6]. ML can also lead to the discovery of novel therapeutics through analyzing
of large biological databases and finding patterns within them. In near future, the
ML model will also support the radiologists as well as physicians and will boost
medical research and practice [7]. Kumari et al. [8] presented the transition of ML
toward cloud toward the fog and IoT-based healthcare. Multan et al. [9] also dis-
cussed growth in fog computing in the healthcare IoT domain. Mohanta et al. [10]
discussed shifting of paradigm toward healthcare using AI, IoT, and 5G technolo-
gies. Rong et al. [11] reviewed some of the applications using epileptic seizures and
filling of a dysfunctional urinary bladder by the use of AI. With increasing avail-
ability of clinically relevant databases, ML is applied to Healthcare 4.0 for diag-
nostic/identification tasks and prediction tasks [12,13]. In recent years, ML is
gaining importance across several disciplines and led to emerging breakthroughs in
Healthcare 4.0 and machine translation [14,15]. In summary, the major contribu-
tions of this chapter are enumerated as follows.

● This work provides an overview of Healthcare 4.0 and the major technologies
used in it.
● This work categorizes and discusses different applications of Healthcare 4.0.
● This work presents an in-depth analysis of ML algorithms with the main focus
on their applications for healthcare.
● This work presents security and other vulnerabilities involved in adopting ML
for healthcare.
 Machine learning for Healthcare 4.0 307

● Finally, this work presents different solutions that can be used for ensuring the
security and efficacy of ML systems in Healthcare 4.0.
The remainder of the chapter is organized as follows. Section 16.2 presents
Healthcare 4.0 and the technologies involved in it, Section 16.3 presents an over-
view of different algorithms involved in ML that are used in healthcare,
Section 16.4 presents ML applicability in healthcare and different functions.
Section 16.5 presents the challenges faced during the deployment of ML in
Healthcare 4.0. Section 16.6 concerns with the solutions to enhance the security for
ML-based systems in Healthcare 4.0. Finally, Section 16.7 concludes the work.

16.2 Healthcare 4.0

When technologies such as blockchain, the Internet of Health Things (IoHT), HC,
big data, medical cyber-physical systems (MCPS), and other smart algorithms are
integrated together to be deployed in order to provide a new vision to the health-care
industry, it is known as Healthcare 4.0. It aims to provide better, cost-effective, and
reliable services to the patients with enhanced efficiency and effectiveness of the
health-care system. The main principles of Healthcare 4.0 are taken from Industry 4.0
[16–20]. Thus, Industry 4.0 inherits the 4.0 version of healthcare; however, there has
been no existence of Healthcare 1.0, 2.0, and 3.0. Healthcare 4.0 in spite of providing
many advantages to the system is complex to build and achieve. These complexities
can be solved by adopting an advanced service-oriented middleware framework. This
will not only help us to integrate different advanced technologies that will be
incorporated with real-time data collections, enhanced AI, and interactive virtual
interfaces facilitating advanced inquisitive solutions but, it will also increase the use
of current services and their platforms. The main aim of Healthcare 4.0 is to create a
health-care value chain with a smart health-care network of patients, hospitals,
clinics, medical devices, and suppliers [21].

16.2.1 Main technologies of Healthcare 4.0

Here is the summarized description of the major technologies used in
Healthcare 4.0.
● Internet of Things (IoT): Also known as the IoHT, which enables the con-
nection of the medical devices in the network [22]. Some examples are virtual
hospitals/wards, insulin pens, wearable biosensors, smart thermometers, low-
cost IoT sensor systems for real-time remote monitoring [23], etc. Qadri et al.
[24] have shown the impact of IoT on the advancement of the health-care
industry.
● Internet of Services (IoS): These medical services and health-care devices are
provided on well-integrated interfaces and software over the Internet [25].
These services are essential for automation and collaboration.
● Health cloud: It is a health information technology customer relationship
management system that incorporates the relationship of doctor–patient and
308 The Internet of medical things

includes record management services. It is the cloud-based technology that

provides data storage, various software resources, and applications, which
requires high storage and rigorous computations [26]. The HC also provides
advantages in cost, security, scalability, and collaboration. For example, cloud-
based health-care monitoring system using IoT [27], cloud-based approach in
designing electronic health records (EHRs), which can be interoperable [28].
Gorp et al. [29] discussed that EHRs should be the lifelong property of patients
and they can share these any time with virtual machines via cloud technology.
Cloud computing with other technologies reduces the cost and also helps in the
timely diagnosis of patients [30]. However, there are certain challenges for
cloud technology to be adopted in healthcare. Idoga et al. [31] discussed some
factors affecting the attitude of health-care consumers toward adopting cloud-
based health centers.
● Health big data analytics: Big data is the huge volume of information on
various digital technologies that collects records of patients. It helps in
managing the hospital’s performance. Big data analytics is used to discover
mechanisms and processes to find the correlations and insights of the data.
Analysis of this big data provides advantages in health-care systems like
enhanced healthcare and cost-effectiveness. Examples of big data in healthcare
are big data for personalized healthcare [32], big data analytics used in opti-
mizing EHRs [33], etc. Li et al. [34] presented a case study of China for
applying big data governance to regional health information networks. Lin
et al. [35] did a survey that targets chronic diseases and health monitoring big
data technologies.
● Medical cyber-physical systems (MCPS): It is integration for a network of
medical devices to the cyber world through computer networks via physical
processes. This helps in fulfilling the vision of continuous availability of high-
quality health-care services [36]. Wireless sensors are critical parts of these
MCPS systems. Some examples of CPS include medical monitoring,
implantable medical devices [37] like, cardiac pacemakers, coronary stents,
implantable insulin pumps, and hip implants. Qiu et al. [38] have proposed
secured data storage and sharing using MCPS system.
● Health fog: HF is basically fog computing which is a mediator layer between
the cloud and the end user that provides on-demand computation, software
resources, and different services of health-care applications [39]. It is efficient
in reducing extra cost incurred for the communication in related systems. It has
the capability of collaborating data from various sources, maintaining proper
security, and privacy concerns. In [40], fog-assisted IoT-enabled patient health
monitoring is presented.
● Mobile communication networks (5G): The 5G communication network
provides a far higher level of performance in terms of communication speed. It
contributes to the health-care system by supporting high-speed data transmis-
sion for health-care applications, real-time sharing of data, facilitating real-
time consultations, and decision-making across the network [41]. Overall, this
would foster health-care transformation and delivery innovations.
 Machine learning for Healthcare 4.0 309

● Blockchain: The concept is used to denote record in the form of blocks and is
very sensitive to any change in the whole chain that is formed using combi-
nation of blocks. Blockchain can facilitate various applications in Healthcare
4.0 such as protecting medical data, sensitive personal patients’ symptoms
stored in the form of data, EHRs data management, etc. Blockchain also pro-
vides security in sharing of data and protection of the patient’s privacy [39,42].
There are numerous applications of blockchain in healthcare. Wang et al. [43]
discussed a framework of parallel health-care systems based on the artificial
systems, including computational experiments together with parallel execution
(ACP) approach. Zarour et al. [44] discussed the impact of blockchain models
for Secure and Trustworthy Electronic Healthcare Records.

16.2.2 Health 4.0 objective

There are many objectives of Healthcare 4.0 but, broadly it can be divided into two.
First one is by enhancing the quality of services with cost effectiveness. Second one
by more efficient and effective systems. Other than this, there are also other ben-
efits like secure data collection and transmission with efficient management,
management of patient’s records, enhanced rate of automation, etc. Healthcare
4.0 has many capabilities within itself, which are when utilized to the full potential
will help to achieve desired abovementioned objectives. This section discusses the
objective that can be achieved in Healthcare 4.0.
● Efficient management of patient’s databases: Healthcare 4.0 involves tech-
nologies like HC, big data, and HF which ease out the management task of the
patient’s record as these are large databases that require intensive administration.
● Less dependency on humans and more on machines: The technologies like
MCPS help in efficient automation in Healthcare 4.0 which provides better
efficiency and accuracy, thus, reducing the reliability of staff or workers.
● Better synchronization and administration of health-care staff/workers
and professionals: Improved resource allocation in health-care staff with
better scheduling by analyzing the current and past data of each professional
will result in a better yielding team in the system and will increase the satis-
faction from the patients.
● The enhanced doctor–patient relationship and customer satisfaction:
Healthcare 4.0 technologies when put together provide people with a better and
efficient overall system in a cost-effective manner. If Healthcare 4.0 is able to
utilize its capabilities to a complete extent, the trust between patient and doctor
relation will increase many folds. As a result, the system will be more reliable
for everyone.
● Improved services in remote areas: Healthcare 4.0 and its technology helps
to make healthcare accessible in the remote areas with virtual appointments
and conferencing with health professionals provided in a cost-effective way.
Also, technologies enable the effective sharing or transmission of highly spe-
cialized resources to remote areas. Thus, expensive health-care resources will
be effectively allocated and shared.
310 The Internet of medical things

● Better accuracy and efficiency of the services: Employing all the technolo-
gies may also help in better maintenance of resources and on-time servicing of
the resources. This will increase the predictability and efficiency of the whole
system.
● Efficiently analyzing patients’ data (medical and nonmedical): Getting a
detailed view of patients’ data that can check other factors responsible for the
disease and can help the professionals in better prediction and creating more
personalized treatments.
● Efficient data analysis: Healthcare 4.0 can help efficient data collection and
transmission on a large scale, these data when analyzed for longer duration
could help in better generalizations as well as results like data of infectious
disease will tell the hotspots of that diseases, age groups affected, moreover it
will help in controlling the rate of infection at large scale.
● Smart decision-making supported by MCPS: Using MCPS in Healthcare
4.0, people can develop smart equipment and automation which can be utilized
to give better prediction and enhanced decision-making. This will also require
less staff and error could be minimized at certain levels, furthermore providing
the people with cost friendly schemes.
● Making a cost-effective system for all: There are many ways to achieve this
in Healthcare 4.0. For example, if people allow proper sharing of advanced,
costly resources and equipment across all health-care providers, the opera-
tional cost will become less and the resources will be utilized in a better way.

16.2.3 Health 4.0 application

This section discusses the applicability of Healthcare 4.0 for patients, for the
health-care professionals, and its applications for resource management [39].
Various applications related to Healthcare 4.0 are discussed in the next subsections.

16.2.3.1 Application for patients

There are various applications that could be for patients in a health-care system.
Few prominent are discussed in coming subsections.
● Management of patient: Today many applications are available for the patients
for various purposes but, having similar functionality, these applications in
Healthcare 4.0 can be integrated together and efficient data analysis can be done,
the patients can use these applications to define their needs which will give the
patient more personalized experience in the system. The huge databases having
patient records can be also managed effectively using technologies like big data
and the HC. These data can thus be accessible to the patients, pharmacies,
laboratories, medical professionals, etc. at any time. Furthermore, the technolo-
gies can enable us to support these functionalities like integrating different sys-
tems. For example, doctors can access patients’ fitness and other activity records
from their personal devices maintaining the integrity and privacy of patients.
● Boosting public healthcare: Healthcare 4.0 enables to boost the maintenance of
overall healthcare of the public in several ways, each way using the technologies
 Machine learning for Healthcare 4.0 311

in a synchronized way. For example, employing ML and DL algorithms inte-

grated within big data, HC, MCPS technologies over large data accumulated
over a period of time can help to predict future outcomes of certain infectious
diseases, health trends, etc. This will help to reduce the vulnerability and risk of
loss of life during a medical pandemic. These technologies will also get us to
know the hotspots of these diseases and these things will result in efficient
management by health-care services.
● Inpatient care: This refers to how Healthcare 4.0 will facilitate the patient
after he/she is admitted to a health-care facility by making the task convenient
for both patient and medical professionals. This includes digital monitoring of
patients, treatment devices, and other applications. These services can be
integrated easily with the given technologies and will provide the patient with
personalized and enhanced health-care experiences such as customized rooms
and meals. Data of a patient can be fetched from other health-care facilities
using network devices and can be employed for better outcomes. Applications
can be used for scheduling medications and lab timings accordingly with
proper alerts. This will help to reduce time loss and saved time can be utilized
efficiently. Similarly, technology like blockchain will help to manage the
transactions of patients efficiently and proper discharge of patients from the
health-care facilities can be done.

16.2.3.2 Application for health-care professionals

There are various applications that could be for health-care professionals. Few
prominent are discussed in coming subsections.
● Efficient management: Management of health-care services by all the medi-
cal professionals is a perplexing task like administration of various types of
doctors, medical staffs, 247 work tasks, and regular scheduling. These all are
the challenges to the efficiency of whole health-care systems. But, Healthcare
4.0 provides various applications and technology that can make this complex
task insanely convenient. Integrating these technologies within themselves will
benefit us first with better collection of data helping in managing operational
procedures, scheduling, alert management, etc. Second, by easily accessible
resources through fast transmission. Third, by better decision-making after
deploying different algorithms. Fourthly, by efficient predictability for future
needs and exploring new trends in healthcare, etc., and finally by deploying
analytical models for providing personalized treatment facilities to each and
every patient. Similarly, there are many more examples. These techniques will
ease out the mundane task of management of the health-care services and will
save a lot of time which can be later used to focus on the patient treatments
more efficiently.
● Better scheduling and resource allocation: A health-care system requires
scheduling in such a way that it can efficiently work 247 shifts, this is a
mundane task, and many other factors related to each medical personnel are
also required to be considered to create a well-optimized schedule. Similarly,
312 The Internet of medical things

health-care professionals require a lot of resources to complete their jobs.

These resources need to be well organized to become available for access at
the required time. To ease out these tasks, Healthcare 4.0 applications can be
designed to work on each medical professional’s data related to resources,
experiences, demands, history, etc. which can be used to put him/her in a
proper schedule. This can help in better optimization and can also help to
manage the emergency conditions. For resource allocation, people can use
transmission of data, resource inventory management to utilize the resources
efficiently.
● Collaboration: Collaboration is required among various health-care systems
in several cases. These can be either within the same health-care facilities or
between diverse health-care facilities and entities. For example, doctors
require consulting with other doctors for some critical cases, connecting with
other facilities for suppliers of pharmaceuticals, medicines, etc. This colla-
boration requires the exchange of data over the systems maintaining security as
well as privacy of data. These are basically the task of administrators in the
health-care system. Healthcare 4.0 provides various applications by combining
different technologies and facilitating easy communications between different
organizations.

16.2.3.3 Application for resource management

Any health-care system largely depends on its resources for its efficient function-
ing. Resources include medical components like medical devices, laboratory tools,
surgical equipment, diagnostic tools, medications, protection gears and instru-
ments, etc. It also includes physical infrastructural tools like a fire alarm, beds,
systems, furniture, water, and power supply. Moreover, all staff, professionals in
healthcare using the abovementioned resources are also resources for the system.
These all resources need to be efficiently utilized and managed in the system. Users
need to consider different aspects for the maintenance of the resources. First, the
availability of resources is an important aspect, which means the resources need to
be available when required. These resources should be equipped with connectivity
like Bluetooth, wired, and wireless. This will help Healthcare 4.0 applications to
link all the resources in order to collect, exchange, and transfer the data from these
resources to the analysis modules. Second, users need proper allocation of resour-
ces to the respective professionals according to the procedures and operations.
These include X-ray machines, magnetic resonance imaging (MRI) scans, ICU
equipment, etc. Thus, there is a need to optimize the allocation of all these
resources using the Healthcare 4.0 applications for scheduling of resources using
data and this can benefit in reducing waiting time and enhanced utilization. Lastly,
the resources must be checked for faults from time to time, and the management of
this fault checking should be properly synchronized with other schedules.

16.2.3.4 Applications and enabling technologies mapping

The chapter has researched till now that Healthcare 4.0 is the integration of various
technologies to serve great objectives for the overall systems. HF is known to
 Machine learning for Healthcare 4.0 313

provide local storage and accessibility of data enabling smooth communication,

while IoS is essential to all the services used by the people or users. Thus, proper
mapping of these technologies is a must. Some technologies like HF and IoS are
almost required in every health-care application. IoT is required for most of the
monitoring devices, blockchain, cloud, and big data for scheduling and collabora-
tion. So, there is a need to properly integrate these technologies so that they can be
utilized to their full potential.

16.3 Machine learning algorithms

ML is based on the concept of learning, training, and then decision-making by
machine based on dataset [45]. This learning can be broadly classified into three
types that are supervised learning (SL), unsupervised learning (UL), and reinfor-
cement learning (RL). SL means it provides the model with labeled data, i.e., the
user supervises the model to learn, while in UL there is no labeled data, the model
learns all by itself. RL is different from these two as it uses reward or punishment-
based approach to making the model learn and predict. Some popular and widely
used algorithms of SL are logistic regression, linear regression (LR), support vector
machine (SVM), decision tree (DT), artificial neural network (ANN), etc., whereas
principal component analysis (PCA) is one of the most popular algorithms of UL.

16.3.1 Linear regression

In LR, the machine models the relationship between independent variables and a
dependent variable (labels and features) and finds a line that best fits the data. To check
closeness of data, the regression line fits the data and users use statistical measures such
as R squared, mean squared error (MSE). A higher R squared value and lower MSE
mean the model fits the data better. The LR technique can be used for three purposes,
determining the strength of predictors, forecasting an effect, and trend forecasting.
Further, it uses cost function and gradient descent techniques to minimize the error.

16.3.2 Logistic regression

This algorithm is used in classification problems that models conditional prob-
ability (0 or 1). In other words, it is similar to the LR model that tries to predict an
outcome for a dependent variable from one or more independent variables but, in
logistic regression, the independent variable can be categorical or continuous. It
uses a logistic function also known as the sigmoid function to classify the data.
Logistic regression is mostly used in binary classification problems. It uses a
decision boundary approach to classify the data into two categories. It also uses
optimization techniques like cost function and gradient descent.

16.3.3 Support vector machines

SVM is another SL algorithm that can be used for both regression and classification
problems. However, it is mostly preferred for classification problems. It can
314 The Internet of medical things

classify both linear and nonlinear data. It works by plotting each data item as a
point in n-dimensional space (where n is the number of features) with the value of
each feature being the value of a particular coordinate. Then, it performs classifi-
cation by finding the hyperplane that differentiates the two classes very well. Data
points are decided on the basis of boundaries which are overlaid by hyperplanes.
Data points that fall on any side of the given plane are attributed to various classes.
Moreover, the dimension highly depends upon the total number of features. For
example, the output would be only one line if in case there are two inputs to
machine. Support vectors are set of data that directs toward the hyperplane and
these will help users to maximize the margin of the classifier. For nonlinear data,
SVM uses kernel functions that take input as low dimension feature space and
outputs high-dimension feature space due to which classification becomes easy.
SVM classifiers like one versus all and one versus one are used for multiclass
classification.

16.3.4 Artificial neural network

These are the algorithms that model similar functioning to the human brain. As
the human brain consists of a network of neurons that are interconnected, ANN
also consists of neurons that are arranged in several layers. There are different
types of ANN such as feedforward neural network (FFNN), recurrent neural
network, and convoluted neural network (CNN). It contains a network of inter-
connected nodes. Each node is a perceptron showing similarity to multiple LRs
which produces signals that are fed into an activation function by the perceptron.
FFNN consists of an input and output layer with a hidden layer within itself. It
feeds the signal in the forwarding direction only. A training algorithm is used by
ANN to minimize the error among targeted and predicted output by adjusting the
weight of each neuron. This training algorithm is known as backpropagation.
Similarly, recurrent neural network (RNN) uses loops in hidden layers due to
which it has internal memory. This memory feature of RNN, in combination to
long short term memory (LSTM) technique (that enhances the memory time), is
used in major technologies today.

16.3.5 Decision tree

DT is a nonparametric supervised algorithm that can be used for both classification
and regression problems. It is basically a flowchart drawn upside down, consisting
of internal nodes, leaf nodes, and a root node. Each internal node represents a test
on a feature, whereas each leaf node represents the class labels and branches
represent conjunctions of features that lead to those class labels. The paths from the
root to leaf represent classification rules. By drawing this flowchart, users can
easily understand the relations. It is also known as classification and regression
trees labels. The paths from the root to leaf represent classification rules. DT is very
useful where concepts for statistics of data are used such as in the case of ML and
data mining. DT falls under domain of SL and it differentiates the data according to
the conditions involved in decision-making process.
 Machine learning for Healthcare 4.0 315

16.3.6 Random forests

Random-forest-SL is a network of DTs. The group of DTs is trained with the bagging
method. In other words, it is a multiple DT that is merged together to give better
predictive results with more accuracy and reliability. It also provides versatility as it
can be used for both classifications and falls under regression problems. It gives
better prediction than DT as it adds more randomness to the trees and during the
splitting of a node, it uses the best feature among a random subset of features. There
is less chance of overfitting in this algorithm, which is a big advantage.

16.3.7 K means
It is an unsupervised algorithm that comes under type of clustering technique.
Clustering in UL is an important approach as it uses patterns in a collection of
uncategorized data and finds natural clusters or groups within the data. K Means is
a subtype of types of clustering and comes under partitioned clustering. It is basi-
cally a centroid-based clustering. K Here represents a number of groups and the
algorithm starts from identifying the k centroids and then iterates between two steps
of assigning the data points and updating these centroids. Euclidean distance
measures the next step for updating centroids by taking the mean of all data points
assigned to a particular cluster.

16.3.8 Naı̈ve Bayes

It is a classification algorithm that uses the Bayes theorem assuming that the pre-
sence of a particular feature in a class is unrelated to the presence of any other
feature. It is easy to build and even works efficiently on large datasets. In spite of its
simplicity, naı̈ve Bayes outperforms many complex sophisticated algorithms.

16.3.9 Dimensionality reduction algorithm

These algorithms are applied to datasets with p dimension of data into subsets of k
dimensions (k<<<n). They serve many benefits in handling and processing large
datasets in ML models such as the space required is reduced, less computational or
training time is required, reduced dimension, which is beneficial to some algo-
rithms, dimension reduction techniques that take care of multicollinearity, reduced
dimensions of datasets, which helps to visualize better, and many more. The
common dimensionality reduction techniques used are feature selection in which
the user only keeps the most relevant variables from the original datasets, while
another technique includes finding a smaller set of new variables, each being a
combination of the input variables, containing basically the same information as the
input variables.

16.3.10 Gradient boosting algorithm and AdaBoosting

algorithm
Boosting algorithms come under ensemble learning techniques. Ensemble learning
increases the accuracy of prediction by combining the decisions by multiple
316 The Internet of medical things

models. This diversification increases the overall performance. Boosting algo-

rithms are very useful for conversion from weak learner to stronger one. It fits a
sequence of weak learner models, which are approximately little stronger to ran-
dom one. Due to their ability to enhance the accuracy of predictions, it is the most
preferred ML algorithm.

16.4 Machine learning for Healthcare 4.0

In the previous section, the chapter has discussed different ML algorithms and the
type of learning it follows; in context to this user will now see how these algorithms
can be utilized to build health-care applications. UL algorithms (learning from
unlabeled data) like clustering can be used for anomaly detection [45]. Some other
examples can be predicting hepatitis using PCA [46], heart diseases can be pre-
dicted using clustering [47], etc. SL includes classification and regression pro-
blems. Moreover, classification techniques are widely used in healthcare for
purposes such as recognizing different body organs for medical images [48], clas-
sifying different lung diseases [49], etc., whereas semi-SL (where the data is both
labeled and unlabeled) techniques are particularly useful for health-care applica-
tions as it is generally difficult to collect sufficient amounts of labeled data. In [50],
clustering based on semi-SL facilitating health-care applications is discussed. Data
sensors used in activity recognition using semi-supervised algorithms are presented
in [51]. In [48,52], medical image segmentation using a semi-supervised approach
is applied by the authors. Apart from these, RL is another learning paradigm in
which the model learns to make a sequence of decisions and the model gets either
rewards or penalties for the actions it performs [6]. RL is known to have great
potential to transform health-care applications. Some uses of RL such as checking
context-free symptom in disease diagnosis are presented in [51].
Further, the healthcare on the basis of using ML can be divided into four
applications: prognosis, diagnosis, treatment, and clinical workflow. Prognosis is
the stage of prediction of the development of disease in patients. In this phase,
symptoms and signs for the disease are identified by predicting if these symptoms
will deteriorate, enhance, or remain steady over a period of time, chances of sur-
vival, etc. ML is used for cancer detecting in brain tumor [53] and lung nodules
[54] are designed to facilitate disease prognosis [55]. These models are tested and
trained on the data collected from patients during this stage. This data comprises
medical images, phenotypic, genotypic, pathology results, etc. ML is useful for
disease prognosis and has been efficiently utilized to achieve the aim of “perso-
nalized medicine” but this field is still budding and many new techniques are yet to
be explored in this. The next application of ML is in diagnosis. Medical image
analysis is one of the methods used for diagnosis. In this ML algorithms are used to
efficiently analyze the medical images for diverse information to be used later in
the treatment phase. There are various imaging appliances like MRI and computed
tomography that provide the medical images of various organs of bodies, which
helps in diagnosing the body for a particular disease, its sign, and symptom.
 Machine learning for Healthcare 4.0 317

Medical image analysis includes detection (identifying specific disease patterns),

classification (identifying different organs and abnormalities), segmentation (seg-
menting different tissues and organs for quantitative analysis), retrieval (to manage
and query huge databases of the present as well as past diagnosis), reconstruction
(raw data from image sensors are used to generate new images), and image regis-
tration (mapping input image to a reference image) tasks. These tasks are efficiently
performed by deploying ML models in designing fully automated intelligent systems.
Today, hospitals and other health-care facilities are producing EHRs in huge
amounts. It comprises structured and unstructured data. ML models have been
designed to extract the clinical features out of these dates which help in improved
diagnosis [56]. In [57], semi-supervised models are designed to extract the data from
unstructured data and they are presented using ML models used for diabetes diag-
nosis [58]. The third application is treatment, in which people can use techniques like
real-time health monitoring using ML, which helps in efficient monitoring of critical
patients, IoT sensors, health tracker smart devices, etc. All the techniques collect
health data of patients, which is communicated over cloud and later analyzed via ML
models, later the predictions are transferred back to originating devices via the cloud
and necessary actions are taken accordingly [59]. Image interpretation of the medical
images by experts is also used in the treatment phase. These interpretations are vital
to prepare clinical reports. Not all health-care service providers are equipped with
expertise like experienced radiologists, technicians. Therefore, making clinical
reports in these cases is challenging. On the other hand, even if hospitals are having
experts with them, making high-quality reports is very time-consuming and trou-
blesome. Natural language processing (NLP) with ML technique can be solutions to
these problems, for example, annotating clinical reports using NLP-based method is
used for such precise analysis [60], an automatic report can be generated by
deploying models that use CNN along with LSTM network [61]. Thorax disease
classification as well as reporting of chest X-rays model can be developed by inte-
grating CNN and RNN [62] and many more. Ahmad et al. [63] have presented
interpretable ML models that are needed in healthcare and discussed, including their
deployed. Figure 16.1 illustrates the different phases of ML in health-care systems.
Deploying different ML models for early prediction and diagnosis of diseases
enables efficient and enhanced treatment for the disease. This is so far one of the best
applications of ML. In a similar study [64], it was concluded that ML techniques
improve efficient predictions by studying ML algorithms that are tested and trained on
clinical data for predicting the risk of cardiovascular disease. Fatima et al. [64] sur-
veyed various ML concepts in order to detect various diseases like liver disease, den-
gue, diabetes. ML model can diagnose as well as predict worsening of cancer [65].
Apart from this, computer-aided detection (CADe) and computer-aided diagnosis
(CADx) systems are developed, which use integrated technologies like image proces-
sing techniques. One such system is IBM’s Watson. As discussed earlier, RL-based
models are also used for diagnosis and treatment [66]. ML models are used in modeling
the clinical time-series data, which help in applications such as prediction of clinical
intervention in ICUs using models with CNN and LSTM [67], managing ICU’s pre-
dictions using attention model [68], and so on. To improve clinical research and
318 The Internet of medical things

Use case

Data access and

anonymization

Data annotation

ML model

Algorithms audit

Real world
survellience

User acceptance

Clinical integration

Regulatory

Multisite validation

Figure 16.1 Phases of deploying ML in healthcare

practice, NLP techniques are used with other medical related software for analyzing
information from unstructured medical data [69]. ML model is also facilitating the
clinicians with applications for the processing of clinical speech as well as audio data,
which helps in saving their time and reducing workload. It also helps in the diagnosis of
disease related to speech such as dementia [70] and Alzheimer’s disease [71].

16.5 Vulnerabilities for ML in Healthcare 4.0

Deploying ML in health-care provides many benefits to the system but creates
many sources of vulnerabilities due to its security and privacy risks. Figure 16.2
depicts various vulnerability sources in deploying ML for healthcare.
 Machine learning for Healthcare 4.0 319

Data
collection

Data
annotation

Different
vulnerabilities Model
trainings

Model
depolyment

Testing

Figure 16.2 Different sources of vulnerability and security threats for ML in

healthcare

16.5.1 Vulnerabilities in data collection

ML models require training and testing of models by using large amounts of data.
The healthcare can use these data in the form of medical image, radiology report,
EHR, etc. that require human efforts in collecting it and it is a time-consuming task.
These data can have various sources for vulnerabilities that might affect ML
implementations. Many times these data collected can have noise (instrumental and
environmental disturbances) in them, which can cause the model to over fit the
data. For example, multi-shot MRI is highly sensitive to motion, thus a small
change in the subject’s head or respiration can cause undesirable noises in images,
which will increase the chances of misdiagnosis. ML/DL system in healthcare
requires qualified technical staff like engineers and data scientists for their efficient
functioning. Since a health-care system is an ecosystem of interdisciplinary tech-
nical and nontechnical staff, it often lacks qualified personnel required for devel-
oping and maintaining the ML/DL systems. This results in inefficient functioning
and the benefits of ML systems cannot be fully utilized. Many times, the hospitals
depend only on their physician or researchers to operate these ML/DL-based sys-
tems considering the clinical usability of these systems as important. But these
people lack the computational experience to develop or operate this system
resulting in adverse effects in the health-care system.
320 The Internet of medical things

16.5.2 Vulnerabilities due to data annotation

Since most of the models used in ML-based systems are supervised ones and these
models require huge labeled data to be trained upon, thus, these labels are assigned to
each sample of data (such as medical images). This process of assigning labels is
known as data annotation and it is an important part of designing an ML model,
which requires trained and expert personnel. But often due to the lack of such
expertise problems such as label leakage, class imbalance, etc. arises leading to
several vulnerabilities. One such vulnerability can be the ambiguity in medical
datasets such as medical images [72] and even the expert physicians can disagree on
precise diagnostic tasks [73]. Some spiteful users can make the situation much more
difficult by disturbing the data which will make the diagnosis more difficult and
complicated. Vulnerability can arise due to improper annotation. Proper guidelines
are given with privacy and legal considerations, which are to be followed by anno-
tating the dataset for life-critical health-care situations [74]. Datasets that are fre-
quently used in healthcare are generally annotated for coarse-grained labels but, in
practical the ML as well as DL model uses fine-grained labels within the clinical
environment. If the label annotation is not done properly, it can lead to problems such
as improper distribution of samples in classes, which can lead to biasing in models,
availability of datasets required, augmentation of datasets, data missing, etc.

16.5.3 Vulnerabilities in model training

A model training (MT) can give targeted outputs efficiently only when it has been
trained properly. Training of an ML model depends on many parameters (such as
learning rates, epochs, and batch size) all of these are needed to be fulfilled.
Incomplete training, privacy breaching, stealing, or model poisoning are some of
the vulnerabilities in the model. MT is at high risk of security threats such as
adversarial attacks [75], model-based attacks [76], data poisoning attacks. In poi-
soning attack, originality and availability of data are deteriorated by poising the
data with the payload. The poisoning modifies the original dataset and produces
anomaly for the boundaries of victimized ML-based model. The attack is carried
out during training phase to illegally gather sensitive information rather than seiz-
ing the whole system. The main objective of this attack is to get those preserved
dataset in order to carry out further more devastating attacks. These vulnerabilities
are required to be checked as it affects the functioning of ML models, thus reducing
their advantages in the health-care system. These models can be deployed in
security-critical applications like biometrics, fingerprint sensors, and in life-critical
applications. Health-care sector tries their best to keep security hygiene that
defends them against adversary in all possible ways. Therefore, the ML-based
healthcare should also overlay themselves on the concept of cyber defense against
hacker.

16.5.4 Vulnerabilities in deployment phase

The deployment phase is crucial as it involves the deployment of ML models in
various functions and operations in the health-care system. This requires huge
 Machine learning for Healthcare 4.0 321

human resources and decisions made by health-care personnel. This deployment

phase also suffers from various vulnerabilities, which is needed to be checked. The
distribution shifts, and incomplete values or data, hold a major problem in
deployment as the data collected has incomplete values which are simply ignored
during analysis. This kind of model leads to the problem of a healthy person
diagnosed with a disease called false positive and a patient having a disease iden-
tified as healthy known as a false negative. Both of these problems are severe and
affect the integrity of the system.

16.5.5 Vulnerabilities in testing phase

The testing phase of an ML model comes after the training phase. If the training of
the model is not done properly, there can be vulnerabilities in the testing phase.
Such as the outcomes of false positive and false negative come from inadequate
data fed to the model in the training phase. Misinterpretation during the testing
phase is also a common issue. The testing phase is the last phase; hence, these
issues can be checked by considering the vulnerabilities of the sections discussed
earlier. At last, the true meaning of ML-powered health-care system is to deploy
ML in critical health-care applications using analytical methods [77].

16.6 ML-based solutions for Healthcare 4.0

This section will discuss the various solutions and applicability of ML to ensure
secure, private, and robust processing.

16.6.1 Privacy preservation

Ensuring the safety of the privacy of the users is necessary for the health-care
system. As discussed, health-care application built on various technologies such as
ML and DL requires huge datasets, these data can be personal to the users, there-
fore, any breach in such data can affect the system with unavoidable consequences.
Preservation of privacy while deploying ML models essentially means that training
the model should not include any additional information regarding the users. The
information/data are basically available as cloud storage; thus, they can be brea-
ched. Data anonymization techniques are used to check data breaches. However, it
is said that even if the data is anonymized, meaningful information can be inferred
from an individual’s private data [78]. There are many efforts addressing this
problem of privacy in ML-powered healthcare, such as the two server model with
three different protocols. There are various methods that are adopted for executing
secured arithmetic processes over a secure multiparty computation-based environ-
ment [79]. Moreover, softmax and sigmoid functions have been proposed as
alternatives for the nonlinear activation functions used in ML models.
Other techniques are also used to preserve privacy such as cryptographic
approaches. It is used for the training and testing of models where encrypted data
are accessed from multiple parties. Cryptographic approaches use methods like
322 The Internet of medical things

homomorphic encryption where, addition, multiplication operations can be done on

the encrypted data which serves as the basis for computing complex functions. The
second method is garbled circuits, which is used in the case where, the private data
of two parties are used to get the results. Using the homophobic encryption and
garbled circuits, some classification techniques such as naı̈ve Bayes, DT, and
hyperplane decision can be used to build cryptographic blocks [80]. Third method
is secret sharing in which assets can be shared among different parties, including
shareholder, having secret data. Lastly, secure processors are also used which
account for the security of sensitive code from the unauthorized elements at higher
levels.

16.6.2 Differential privacy

It refers to the process of adding disturbances to the dataset in order to protect that
data. This idea was first proposed in 2006 [81], it guarantees the privacy of the
algorithm, which analyses aggregate datasets [82]. The group privacy and sturdi-
ness additional data, which serves many benefits to the health-care applications,
uses the differential privacy (DP) concept. Group privacy is useful when the dataset
contains correlated samples and composability is used for modularity of algorithm
designed. Moreover, this archives robustness without hindering the system’s priv-
acy while using additional information/data. The model for healthcare could also be
developed by working upon the available encrypted datasets that usually have noise
involved in them [83]. Private ML designed by using Private Aggregation of
Teacher Ensembles is also very useful for such system [84]. Other privacy
mechanisms are differentially private stochastic gradient descent [85], hyper
parameter selection [86], Laplace [87], and Exponential Noise Differential Privacy
Mechanisms [88,89], which are beneficial for health-care system.

16.6.3 Federated learning

Google Inc. recently proposed the ideas of federated learning (FL), also known as
collaborative learning, which is a learning technique that the model has trained an
algorithm using local data or data from multiple decentralized servers but, without
any exchange of data [90]. Ye et al. [91] proposed edge-FL, which separates the
process of updating the local model that is supposed to be completed independently
by mobile devices.

16.7 Conclusion
This chapter worked on a way to demonstrate the role of different technologies,
keeping the main focus on ML for Healthcare 4.0. Initially, the chapter defines
Healthcare 4.0 and its applications. Then, various types of ML algorithms are
highlighted. Further, it explores ML applications in healthcare, and their way of
implementation. The chapter also discussed the challenges faced while deploying
ML models in the health-care system. The chapter provides some solutions to
 Machine learning for Healthcare 4.0 323

ensure the security, privacy, and robustness of ML models when implemented over
health-care system. ML is a vast field within itself and when this field is integrated
with other technologies, it produces great results. When used in the health-care
system, this field will boost efficiency and enhance its service to many subfields of
healthcare which consequently benefits the human beings.

References
[1] C. Kanchanachitra, M. Lindelow, T. Johnston et al., “Human resources for
health in southeast Asia: Shortages, distributional challenges, and interna-
tional trade in health services,” Lancet, vol. 377, no. 9767, pp. 769–781,
2011.
[2] M. Akay and T. Tamura, “Global healthcare: Advances and challenge
[Scanning the Issue],” Proc. IEEE, vol. 103, no. 2, pp. 147–149, 2015.
[3] G. Alfian, M. Syafrudin, M. Ijaz, M. Syaekhoni, N. Fitriyani, and J. Rhee,
“A personalized HEALTHCARE monitoring system for diabetic patients by
UTILIZING BLE-based sensors and real-time data processing,” Sensors,
vol. 18, no. 7, p. 2183, 2018. doi:10.3390/s18072183.
[4] M. B. Sampa, M. R. Hoque, R. Islam et al., “Redesigning portable health
clinic platform as a remote healthcare system to tackle COVID-19 pandemic
situation in unreached communities,” Int. J. Environ. Res. Public Health,
vol. 17, no. 13, p. 4709, 2020.
[5] M. Jindal, J. Gupta, and B. Bhushan, “Machine learning methods for IoT and
their future applications,” in 2019 International Conference on Computing,
Communication, and Intelligent Systems (ICCCIS), 2019. doi:10.1109/
icccis48478.2019.8974551.
[6] R. Tayal and A. Shankar, “Learning and predicting diabetes data sets using
semi-supervised learning,” in 2020 10th International Conference on Cloud
Computing, Data Science & Engineering (Confluence), Noida, India, 2020,
pp. 385–389. doi:10.1109/Confluence47617.2020.9058276.
[7] L. Xing, E. A. Krupinski, and J. Cai, “Artificial intelligence will soon change
the landscape of medical physics research and practice,” Med. Phys., vol. 45,
no. 5, pp. 1791–1793, 2018.
[8] A. Kumari, S. Tanwar, S. Tyagi, and N. Kumar, “Fog computing for
Healthcare 4.0 environment: Opportunities and challenges,” Comput. Electr.
Eng., vol. 72, pp. 1–13, 2018.
[9] A. A. Mutlag, M. K. A. Ghani, N. Arunkumar, M. A. Mohammed, and
O. Mohd, “Enabling technologies for fog computing in healthcare IoT systems,”
Future Gener. Comput. Syst., vol. 90, pp. 62–78, 2019, [online] Available:
http://www.sciencedirect.com/science/article/pii/S0167739X18314006.
[10] B. Mohanta, P. Das, and S. Patnaik, “Healthcare 5.0: A paradigm shift in
digital healthcare system using artificial intelligence, IOT and 5G commu-
nication,” in 2019 International Conference on Applied Machine Learning
(ICAML), 2019, pp. 191–196. doi:10.1109/ICAML48257.2019.00044.
324 The Internet of medical things

[11] G. Rong, A. Mendez, E. B. Assi, B. Zhao, and M. Sawan, “Artificial intel-

ligence in healthcare: Review and prediction case studies,” Engineering,
vol. 6, no. 3, pp. 291–301, 2020, [online] Available: http://www.science-
direct.com/science/article/pii/S2095809919301535.
[12] J. Gaur, A. K. Goel, A. Rose, and B. Bhushan, “Emerging trends in machine
learning,” in 2019 2nd International Conference on Intelligent Computing,
Instrumentation and Control Technologies (ICICICT), 2019. doi:10.1109/
icicict46008.2019.8993192.
[13] S. Kumar, B. Bhusan, D. Singh, and D. K. Choubey, “Classification of
diabetes using deep learning,” in 2020 International Conference on
Communication and Signal Processing (ICCSP), 2020. doi:10.1109/
iccsp48568.2020.9182293.
[14] S. Goyal, N. Sharma, B. Bhushan, A. Shankar, and M. Sagayam, “IoT
enabled technology in secured healthcare: Applications, challenges and
future directions,” in A. E. Hassanien, A. Khamparia, D. Gupta, K. Shankar,
and A. Slowik (Eds.), Cognitive Internet of Medical Things for Smart
Healthcare Studies in Systems, Decision and Control (Vol. 311), 2021.
Cham: Springer. doi:10.1007/978-3-030-55833-8_2.
[15] N. Sharma, I. Kaushik, B. Bhushan, S. Gautam, and A. Khamparia,
“Applicability of WSN and biometric models in the field of healthcare,” in
Deep Learning Strategies for Security Enhancement in Wireless Sensor
Networks Advances in Information Security, Privacy, and Ethics, 2020,
pp. 304–329. USA: IGI Global. doi:10.4018/978-1-7998-5068-7.ch016.
[16] C. Thuemmler and C. Bai, Health 4.0: How Virtualization and Big Data are
Revolutionizing Healthcare. New York, NY, USA: Springer, 2017.
[17] N. Mohamed and J. Al-Jaroodi, “The impact of Industry 4.0 on healthcare sys-
tem engineering,” in Proc. IEEE Int. Syst. Conf. (SysCon), Apr. 2019, pp. 1–7.
[18] E. AbuKhousa, N. Mohamed, and J. Al-Jaroodi, “Health 4.0: Can healthcare
industry get smarter?” in Proc. 13th Int. Symp. Tools Methods Competitive
Eng. (TMCE), 2020.
[19] A. Khelassi, V. V. Estrela, A. C. B. Monteiro, R. P. França, Y. Iano, and
N. Razmjooy, “Health 4.0: Applications management technologies and
review,” Med. Technol. J., vol. 2, no. 4, pp. 262–276, 2019.
[20] J. Chanchaichujit, A. Tan, F. Meng, and S. Eaimkhong, “An introduction to
Healthcare 4.0,” in Healthcare 4, Singapore: Palgrave Pivot, pp. 1–15, 2019.
[21] J. Al-Jaroodi, N. Mohamed, and E. Abukhousa, “Health 4.0: On the way to
realizing the healthcare of the future,” IEEE Access, vol. 8, pp. 211189–
211210, 2020. doi:10.1109/ACCESS.2020.3038858.
[22] S. M. R. Islam, D. Kwak, M. H. Kabir, M. Hossain, and K.-S. Kwak, “The
Internet of Things for health care: A comprehensive survey,” IEEE Access,
vol. 3, pp. 678–708, 2015.
[23] M. D’Aloia, A. Longo, G. Guadagno, et al., “Low cost IoT sensor system for
real-time remote monitoring,” in 2020 IEEE International Workshop on
Metrology for Industry 4.0 & IoT, 2020, pp. 576–580. doi:10.1109/
MetroInd4.0IoT48571.2020.9138251.
 Machine learning for Healthcare 4.0 325

[24] Y. A. Qadri, A. Nauman, Y. B. Zikria, A. V. Vasilakos, and S. W. Kim, “The

future of healthcare Internet of Things: A survey of emerging technologies,”
IEEE Commun. Surveys Tutorials, vol. 22, no. 2, pp. 1121–1167, Second
Quarter 2020. doi:10.1109/COMST.2020.2973314.
[25] J. Cardoso, K. Voigt, and M. Winkler, “Service engineering for the Internet
of services,” in Proc. Int. Conf. Enterprise Inf. Syst. Berlin, Germany:
Springer, 2008, pp. 15–27.
[26] E. AbuKhousa, N. Mohamed, and J. Al-Jaroodi, “E-health cloud: Opportunities
and challenges,” Future Internet, vol. 4, no. 3, pp. 621–645, 2012.
[27] H. B. Aziz, S. Sharmin, and T. Ahammad, “Cloud based remote healthcare
monitoring system using IoT,” in 2019 International Conference on
Sustainable Technologies for Industry 4.0 (STI), 2019, pp. 1–5. doi:10.1109/
STI47673.2019.9068029.
[28] A. Bahga and V. K. Madisetti, “A cloud-based approach for interoperable
electronic health records (EHRs),” IEEE J. Biomed. Health Inf., vol. 17,
no. 5, pp. 894–906, 2013. doi:10.1109/JBHI.2013.2257818.
[29] P. Van Gorp and M. Comuzzi, “Lifelong personal health data and applica-
tion software via virtual machines in the cloud,” IEEE J. Biomed. Health
Inf., vol. 18, no. 1, pp. 36–45, 2014. doi:10.1109/JBHI.2013.2257821.
[30] C. Xu, N. Wang, L. Zhu, K. Sharif, and C. Zhang, “Achieving searchable
and privacy-preserving data sharing for cloud-assisted E-healthcare system,”
IEEE Internet Things J., vol. 6, no. 5, pp. 8345–8356, 2019. doi:10.1109/
JIOT.2019.2917186.
[31] P. E. Idoga, M. Toycan, H. Nadiri, and E. Çelebi, “Factors affecting the
successful adoption of e-health cloud based health system from healthcare
consumers’ perspective,” IEEE Access, vol. 6, pp. 71216–71228, 2018.
doi:10.1109/ACCESS.2018.2881489.
[32] M. Viceconti, P. Hunter, and R. Hose, “Big data, big knowledge: Big data
for personalized healthcare,” IEEE J. Biomed. Health Inf., vol. 19, no. 4,
pp. 1209–1215, 2015. doi:10.1109/JBHI.2015.2406883.
[33] C. Zhang, R. Ma, S. Sun, Y. Li, Y. Wang, and Z. Yan, “Optimizing the
electronic health records through big data analytics: A knowledge-based
view,” IEEE Access, vol. 7, pp. 136223–136231, 2019. doi:10.1109/
ACCESS.2019.2939158.
[34] Q. Li, L. Lan, N. Zeng, et al., “A framework for big data governance to
advance RHINs: A case study of China,” IEEE Access, vol. 7, pp. 50330–
50338, 2019. doi:10.1109/ACCESS.2019.2910838.
[35] R. Lin, Z. Ye, H. Wang, and B. Wu, “Chronic diseases and health mon-
itoring big data: A survey,” IEEE Rev. Biomed. Eng., vol. 11, pp. 275–288,
2018. doi:10.1109/RBME.2018.2829704.
[36] I. Lee, O. Sokolsky, S. Chen et al., “Challenges and research directions in
medical cyber–physical systems,” Proc. IEEE, vol. 100, no. 1, pp. 75–90, 2012.
[37] J. A. Hansen and N. M. Hansen, “A taxonomy of vulnerabilities in
implantable medical devices,” in Proc. 2nd Annu. Workshop Secur. Privacy
Med. Home-Care Syst., 2010, pp. 13–20.
326 The Internet of medical things

[38] H. Qiu, M. Qiu, M. Liu, and G. Memmi, “Secure health data sharing for
medical cyber-physical systems for Healthcare 4.0,” IEEE J. Biomed. Health
Inf., vol. 24, no. 9, pp. 2499–2505, 2020. doi:10.1109/JBHI.2020.2973467.
[39] J. Al-Jaroodi and N. Mohamed, “Blockchain in industries: A survey,” IEEE
Access, vol. 7, pp. 36500–36515, 2019.
[40] P. Verma and S. K. Sood, “Fog assisted-IoT enabled patient health mon-
itoring in smart homes,” IEEE Internet Things J., vol. 5, no. 3, pp. 1789–
1796, 2018. doi:10.1109/JIOT.2018.2803201.
[41] D. M. West, How 5G Technology Enables the Health Internet of Things,
Brookings Center for Technol. Innovation, vol. 3, pp. 1–20, 2016.
[42] N. Mohamed and J. Al-Jaroodi, “Applying blockchain in industry 4.0
applications,” in Proc. IEEE 9th Annu. Comput. Commun. Workshop Conf.
(CCWC), 2019, pp. 0852–0858.
[43] S. Wang, J. Wang, X. Wang, et al., “Blockchain-powered parallel healthcare
systems based on the ACP approach,” IEEE Trans. Comput. Soc. Syst.,
vol. 5, no. 4, pp. 942–950, 2018. doi:10.1109/TCSS.2018.2865526.
[44] M. Zarour, M. T. Ansari, M. Alenezi, et al., “Evaluating the impact of
Blockchain Models for secure and trustworthy electronic healthcare
records,” IEEE Access, vol. 8, pp. 157959–157973, 2020. doi:10.1109/
ACCESS.2020.3019829.
[45] V. Chandola, A. Banerjee, and V. Kumar, “Anomaly detection: A survey,”
ACM Comput. Surv. (CSUR), vol. 41, no. 3, p. 15, 2009.
[46] K. Polat and S. Gunes, “Prediction of hepatitis disease based on principal
component analysis and artificial immune recognition system,” Appl. Math.
Comput., vol. 189, no. 2, pp. 1282–1291, 2007.
[47] A. K. Pandey, P. Pandey, K. Jaiswal, and A. K. Sen, “Data Mining clustering
techniques in the prediction of heart disease using attribute selection
method,” Heart Disease, vol. 14, pp. 16–17, 2013.
[48] D. Mahapatra, “Semi-supervised learning and graph cuts for consensus based
medical image segmentation,” Pattern Recognit., vol. 63, pp. 700–709,
2017.
[49] W. Shen, M. Zhou, F. Yang, C. Yang, and J. Tian, “Multi-scale convolutional
neural networks for lung nodule classification,” in International Conference
on Information Processing in Medical Imaging. Springer, 2015, pp. 588–599.
[50] M. N. Sohail, J. Ren, and M. Uba Muhammad, “A Euclidean group assess-
ment on semi-supervised clustering for healthcare clinical implications
based on real-life data,” Int. J. Environ. Public Health, vol. 16, no. 9,
p. 1581, 2019.
[51] H.-C. Kao, K.-F. Tang, and E. Y. Chang, “Context-aware symptom checking
for disease diagnosis using hierarchical reinforcement learning,” in Thirty-
Second AAAI Conference on Artificial Intelligence, 2018.
[52] W. Bai, O. Oktay, M. Sinclair et al., “Semi- supervised learning for network-
based cardiac MR image segmentation,” in International Conference on
Medical Image Computing and Computer-Assisted Intervention. Springer,
2017, pp. 253–260.
 Machine learning for Healthcare 4.0 327

[53] P. Afshar, A. Mohammadi, and K. N. Plataniotis, “Brain tumor type classi-

fication via capsule networks,” in 2018 25th IEEE International Conference
on Image Processing (ICIP). IEEE, 2018, pp. 3129–3133.
[54] W. Zhu, C. Liu, W. Fan, and X. Xie, “DeepLung: Deep 3d dual path nets for
automated pulmonary nodule detection and classification,” in 2018 IEEE
Winter Conference on Applications of Computer Vision (WACV). IEEE,
2018, pp. 673–681.
[55] A. Collins and Y. Yao, “Machine learning approaches: Data integration for
disease prediction and prognosis,” in Applied Computational Genomics.
Springer, 2018, pp. 137–141.
[56] P. B. Jensen, L. J. Jensen, and S. Brunak, “Mining electronic health records:
Towards better research applications and clinical care,” Nat. Rev. Genet.,
vol. 13, no. 6, p. 395, 2012.
[57] Z. Wang, A. D. Shah, A. R. Tate, S. Denaxas, J. Shawe-Taylor, and H.
Hemingway, “Extracting diagnoses and investigation results from unstruc-
tured text in electronic health records by semi-supervised machine learning,”
PLoS One, vol. 7, no. 1, p. e30412, 2012.
[58] T. Zheng, W. Xie, L. Xu et al., “A machine learning-based framework to
identify type 2 diabetes through electronic health records,” Int. J. Med. Inf.,
vol. 97, pp. 120–127, 2017.
[59] V. Jindal, “Integrating mobile and cloud for PPG signal selection to monitor
heart rate during intensive physical exercise,” in Proceedings of the
International Conference on Mobile Software Engineering and Systems.
ACM, 2016, pp. 36–37. Selection method, heart disease, vol. 14, pp. 16–17,
2013.
[60] J. Zech, M. Pain, J. Titano et al., “Natural language-based machine learning
models for the annotation of clinical radiology reports,” Radiology, vol. 287,
no. 2, pp. 570–580, 2018.
[61] Y. Xue, T. Xu, L. R. Long et al., “Multimodal recurrent model with attention
for automated radiology report generation,” in International Conference on
Medical Image Computing and Computer-Assisted Intervention. Springer,
2018, pp. 457–466.
[62] X. Wang, Y. Peng, L. Lu, Z. Lu, and R. M. Summers, “TieNet: Text-image
embedding network for common thorax disease classification and reporting
in chest x-rays,” in Proceedings of the IEEE Conference on Computer Vision
and Pattern Recognition, 2018, pp. 9049–9058.
[63] M. A. Ahmad, A. Teredesai, and C. Eckert, “Interpretable Machine
Learning in Healthcare,” in 2018 IEEE International Conference on
Healthcare Informatics (ICHI), 2018, pp. 447–447. doi:10.1109/ICHI.
2018.00095.
[64] M. Fatima and M. Pasha, “Survey of machine learning algorithms for disease
diagnostic,” J. Intell. Learn. Syst. Appl., vol. 9, no. 01, p. 1, 2017.
[65] J. A. Cruz and D. S. Wishart, “Applications of machine learning in cancer
prediction and prognosis,” Cancer Inf., vol. 2, p. 117693510600200030,
2006.
328 The Internet of medical things

[66] Z. Zhang, “Reinforcement learning in clinical medicine: A method to opti-

mize dynamic treatment regime over time,” Ann. Transl. Med., vol. 7, no.
14, p. 345, 2019. doi:10.21037/atm.2019.06.75.
[67] H. Suresh, “Clinical event prediction and understanding with deep neural
networks,” Ph.D. dissertation, Massachusetts Institute of Technology, 2017.
[68] H. Song, D. Rajan, J. J. Thiagarajan, and A. Spanias, “Attend and diagnose:
Clinical time series analysis using attention models,” in Thirty-Second AAAI
Conference on Artificial Intelligence, 2018.
[69] A. Névéol, H. Dalianis, S. Velupillai, G. Savova, and P. Zweigenbaum ́ ,
“Clinical natural language processing in languages other than English:
Opportunities and challenges,” J. Biomed. Semant., vol. 9, no. 1, p. 12, 2018.
[70] C. Pou-Prom and F. Rudzicz, “Learning multiview embeddings for assessing
dementia,” in Proceedings of the 2018 Conference on Empirical Methods in
Natural Language Processing, 2018, pp. 2812–2817.
[71] K. C. Fraser, J. A. Meltzer, and F. Rudzicz, “Linguistic features identify
Alzheimer’s disease in narrative speech,” J. Alzheimer’s Dis., vol. 49, no. 2,
pp. 407–422, 2016.
[72] A. Zahin, L. T. Tan, and R. Q. Hu, “Sensor-based human activity recognition
for smart healthcare: A semi-supervised machine learning,” in The International
Conference on Artificial Intelligence for Communications and Networks.
Springer, 2019, pp. 450–472. https://doi.org/10.1007/978-3-030-22971-9_39.
[73] X. A. Li, A. Tai, D. W. Arthur et al., “Variability of target and normal
structure delineation for breast cancer radiotherapy: An RTOG multi-
institutional and multi observer study,” Int. J. Radiat. Oncol. Biol. Phys.,
vol. 73, no. 3, pp. 944–951, 2009.
[74] F. Xia and M. Yetisgen-Yildiz, “Clinical corpus annotation: Challenges and
strategies,” in Proceedings of the Third Workshop on Building and
Evaluating Resources for Biomedical Text Mining (BioTxtM 2012) in con-
junction with the International Conference on Language Resources and
Evaluation (LREC), Istanbul, Turkey, 2012.
[75] C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. Goodfel-low,
and R. Fergus, “Intriguing properties of neural networks,” arXiv preprint
arXiv:1312.6199, 2013.
[76] B. Biggio, B. Nelson, and P. Laskov, “Poisoning attacks against support
vector machines,” in 29th International Conference on Machine Learning,
2012, pp. 1807–1814.
[77] T. J. Pollard, I. Chen, J. Wiens et al., “Turning the crank for machine
learning: Ease, at what expense?,” Lancet Digital Health, vol. 1, no. 5,
pp. e198–e199, 2019.
[78] A. Narayanan and V. Shmatikov, Robust De-Anonymization of Large
Datasets (How to Break Anonymity of the Netflix Prize Dataset), University
of Texas at Austin, 2008.
[79] P. Mohassel and Y. Zhang, “SecureML: A system for scalable privacy-
preserving machine learning,” in 2017 IEEE Symposium on Security and
Privacy (SP). IEEE, 2017, pp. 19–38.
 Machine learning for Healthcare 4.0 329

[80] R. Bost, R. A. Popa, S. Tu, and S. Goldwasser, “Machine learning classifi-

cation over encrypted data,” in NDSS, vol. 4324, 2015, p. 4325.
[81] C. Dwork, Differential privacy. In H. C. A van Tilborg and S. Jajodia (Eds.),
Encyclopedia of Cryptography and Security. Boston, MA: Springer. https://
doi.org/10.1007/978-1-4419-5906-5_752.
[82] M. Abadi, A. Chu, I. Goodfellow et al., “Deep learning with differential
privacy,” in Proceedings of the 2016 ACM SIGSAC Conference on
Computer and Communications Security. ACM, 2016, pp. 308–318.
[83] M. McDermott, S. Wang, N. Marinsek, R. Ranganath, M. Ghassemi, and
L. Foschini, “Reproducibility in machine learning for health,” in Presented
at the International Conference on Learning Representative (ICLR) 2019,
Reproducibility in Machine Learning Workshop, 2019.
[84] N. Papernot, S. Song, I. Mironov, A. Raghunathan, K. Talwar, and
U. Erlingsson, “Scalable private learning with pate,” ́ in International
Conference on Learning Representations (ICLR), 2018.
[85] Y.-X. Wang, B. Balle, and S. Kasiviswanathan, “Subsampled Rényi differ-
ential privacy and analytical moments accountant,” arXiv preprint
arXiv:1808.00087, 2018.
[86] H. B. McMahan, G. Andrew, U. Erlingsson et al., “A general approach to
adding differential privacy to iterative training procedures,” in NeurIPS
2018 Workshop on Privacy Preserving Machine Learning, 2018.
[87] N. Phan, X. Wu, H. Hu, and D. Dou, “Adaptive Laplace mechanism:
Differential privacy preservation in deep learning,” in 2017 IEEE
International Conference on Data Mining (ICDM). IEEE, 2017,
pp. 385–394.
[88] F. McSherry and K. Talwar, “Mechanism design via differential privacy.” in
FOCS, vol. 7, 2007, pp. 94–103.
[89] C. Dwork and F. D. McSherry, “Exponential noise distribution to optimize
database privacy and output utility,” Jul. 14 2009, US Patent 7,562,071.
[90] H. B. McMahan, E. Moore, D. Ramage et al., “Communication-efficient
learning of deep networks from decentralized data,” in Proceedings of the
20th International Conference on Artificial Intelligence and Statistics
(AISTATS) JMLR. WCP, vol. 54, 2017.
[91] Y. Ye, S. Li, F. Liu, Y. Tang, and W. Hu, “EdgeFed: Optimized federated
learning based on edge computing,” IEEE Access, vol. 8, pp. 209191–
209198, 2020. doi:10.1109/ACCESS.2020.3038287.
 Chapter 17
Big data analytics and data mining for
healthcare and smart city applications
Sohit Kummar1 and Bharat Bhushan1

The unprecedented growth of population in urban areas has been causing a chal-
lenge for the citizens in their day-to-day lives such as road congestion, public
security, environmental pollution, electricity shortage and water shortage. To
control and resolve all these issues, new technologies have been developed for
smart cities. Intelligent services and better applications are deployed in smart cities,
by combining the Internet of Things (IoT) with the technologies like data mining
(DM) and deep learning (DL). Many sectors like healthcare, governance, agri-
culture and public safety can increase their efficiency with the help of these new
technologies and can convert these into smart applications for smart cities.
Different kinds of computing like edge computing, fog computing and cloud
computing support to provide better insights into analytics with the help of big data
in smart cities. All these technologies are transforming or raising healthcare eco-
systems, leading them in the direction of smart healthcare. This permits surgeons to
get real-time data of their patients distantly with the help of wireless communica-
tion. Smart healthcare is established on new technologies to convey enriched and
valued healthcare facilities for patients. This chapter explores the current chal-
lenges that are faced during the indigenous development of the smart cities.
Furthermore, the chapter discusses the theoretical background of smart cities with
the explanation of their components. Moreover, the chapter describes the necessity
of computational infrastructure for smart cities in a framework of big data and DM.
The chapter highlights some mining methods for extracting important information
from huge and mixed data. Additionally, the chapter examines the advancement of
healthcare sector in smart cities in context of big data and DM.

17.1 Introduction

Today almost 55% of the population are living in the urban cities, and it is calcu-
lated that in the coming 30 years about 68% of the population will be living in

1
School of Engineering and Technology (SET), Sharda University, Greater Noida, India
332 The Internet of medical things

smart cities [1]. According to an estimation by the United Nations, it is said that the
population will increase by 2.5 billion by the year 2050. This huge amount of
population will pose a great challenge, as it is highly difficult for the development
and sustainable management of the urban cities areas [2]. It will also be a great
challenge for the people to provide a very good quality of life (QoL) for the citizens
of smart cities [3]. Therefore, it is very important to understand the need for urgent
and effective solutions to support various demands of the population that grow at a
very fast speed [4]. On the other side, the evolutions of the big data analytics
(BDA) and the IoT [5] are the major factors that are responsible for the highly
efficient implementation of the smart cities’ services [6]. The BDA and IoT have
gained immense popularity and attention as shown in Figure 17.1.
Across various domains, various intelligent systems have been developed with the
emerging technologies [7], which help us to make different domains that are highly
efficient. The benefit of smart cities is the processing and collection of data by a
different technology of computing, to provide better interconnection and communica-
tion [8]. This leads to a better service in the smart services by uplifting the security,
privacy, availability, sink and transportation. After looking at these issues, to support
the development of the smart cities, authorities have shown a great interest toward the
smart cities and invested billions of dollars [9]. IoT can be referred to as making the
interconnection between the infinite numbers of the smart devices [10]. This technol-
ogy helps the physical objects to enact as human and does their behaviors such as
decision-making, thinking, hearing, sensing and seeing in order to coordinate actions,
share information and communicate with other [11]. By these abilities, one can trans-
form these normally simple objects or physical devices into intelligent devices that can
operate on their own and adjust to the real-time circumstances and work peacefully
without any disturbance of a third member like humans [12].
BDA and DM concepts are specific to smart cities as a whole and especially to
smart healthcare systems in many ways [13]. Social media is reconstructing the health-
related communications and share medical information, by varying the way of keeping
the health records that evaluate by specialists. The availability of web resources on
social media feeds and online forums delivers very important facts [14]. Indeed, BDA

100 Internet
of things
80 Big data
analytics

60

40

20

0
1 9 5 1 9 5 1
0–0 1–0 3–0 5–0 6–0 8–0 0–0
201 201 201 201 201 201 202

Figure 17.1 Google search trends for BDA and IoT

 Big data analytics and data mining 333

and DM techniques, broadcast management schemes look at people’s real-life situa-

tions, producing large amounts of formal and informal information due to the rapid and
massive market of human devices and the development of wireless sensors and mobile
communication technologies [15]. In addition, clinical trials, photographs and
descriptions from clinics lead to patients’ medical supplies, compiled by records that
can take several forms and programs. The most common are electronic health records
(EHR), electrical medical records (EMR) and personal health records (PHR). While the
EHR is expected to report care incidents in most management delivery management,
EMRs of real-time patient health records are validated in decision-making tools, which
can be used to assist physicians in decision-making [16]. PHR is a complete tool; a
lifelong tool for dealing with health-related data, promoting healthcare and supporting
the management of chronic diseases unlike EHR and EMR and is often managed by
patients. Moreover, the methodical static texts reveal an important cause of environ-
mental familiarity [17]. It needs web tools in dedicated mines to hold and refine
recovery. Finally, a large number of biological data (genomics, macrobiotic, pro-
teomics, metabolomics, epigenetics and transcriptomics) are produced and invented at
an unparalleled rate, as the cost of obtaining and testing data decreases with these
technologies [18]. A summary contribution of this work is enumerated as follows:
● This chapter highlights various requirements and explores the characteristic
features of a smart city.
● This chapter describes different kinds of requirements for cloud, fog and edge
computing.
● This chapter presents a brief description of the use of new technology like DM
and DL for smart cities.
● This chapter discusses the environmental sustainability, public safety, health
and urban land planning for smart cities.
● This chapter explores the role of BDA and DM in the healthcare sector.
The remainder of the chapter is organized as follows. Section 17.2 presents the
theoretical background knowledge of the smart cities, in which all the smart
functions of the smart cities highlights. Section 17.3 describes the function of
computational infrastructure to provide the analytics in smart cities along with
overall functioning of cloud computing, fog computing, edge computing and big
data concepts based on machine learning (ML). Section 17.4 provides the insightful
knowledge of mining methods that are available for big data to mine heterogeneous
data. Section 17.5 highlights the role of big data in designing and planning of
healthcare sector of smart cities. Finally, the chapter concludes in Section 17.6 with
some important future research directions.

17.2 Theoretical background of smart cities

Smart cities can be termed as efficient management of the economy, environment,
mobility, living, people and government with the help of the latest technologies. It is
deeply studied and surveyed different aspects of smart city themes and concepts [19].
334 The Internet of medical things

Lazaroiu et al. [20] investigated the different aspects of energy consumption and energy
efficacy problems and were aware of the fact that cities are consuming 75% of the
world’s energy production capacities, and these cities are also responsible for generating
80% of carbon dioxide emissions. For a balanced use of the world’s precious energy
resources, a model is developed that helped us to resolve the issues with energy effi-
ciency. Some components of smart cities are discussed in the succeeding subsection [21].

17.2.1 Smart people

The two main aims for the smart cities are to improve the QoL and a better living
environment, as services and smart devices are mainly used by the users, which makes
it crucial to be well designed and properly implemented for the people [22]. Yeh [23]
suggested that the people of the smart cities should be aware of the social networking
platforms, which work as a bridge for learning the purpose of the efficient use of the
services and systems of smart cities. Belanche-Gracia et al. [24] proposed the security
and privacy concerns for the use of transportation and public facilities with the use of
the advanced smart card with the help of radio-frequency identification technology.
Chatterjee and Kar [25] discussed the factors that affect the use of information systems
based on the class of systems and information. For a mutual living experience, people
should exchange their thoughts and opinions on social media platforms. It also gives an
immense opportunity for people to earn money by resolving resident’s issues by using
these social media platforms. All these data are to be collected and processed; for
example, Niforatos et al. [26] described a crowdsourcing weather application that takes
a mixture of data from manual input of people and sensors. Mone [27] suggested that it
is very crucial to design the smart cities keeping in mind the analyses of traffic issues.

17.2.2 Smart economy

A smart economy can be defined as cities, which uses smart payment tools and
smart money to uplift its economy [28]. Smart economy depends on two terms that
are mobile commerce and smart business, and these two main terms are mainly
responsible for providing smart economy in the smart cities. Johnson et al. [29]
proposed a probabilistic and predictive architecture that depends on management
and risk-based approach to shift the current situations of market within business.
Moreover, commercial and potential benefits of smart economy are based on the
interaction of smart citizens, and most of the smart cities do not have a smart and
good economy due to lack of proper designing and planning. Keegan et al. [30]
described the need for E-commerce in digital cities without creating any problem
for the retailers without hampered their business. To uplift the smart economy of
both, the E-commerce and the retailers should work in collaboration with each
other by providing a platform to ease the business of the smart cities [31]. Mobile
shopping system helps a retailer in the smart cities to know what a customer wants,
which type of smartphone, which in turn helps the retailer to raise the graph of their
sale. However, an equal balance should be maintained between the experience and
innovation of users in terms of the smart economy, while the user’s account privacy
remains a challenge [32].
 Big data analytics and data mining 335

17.2.3 Smart governance

For smart governance, the government should use social media to encourage the
residents to collaborate and participate in the smart cities more efficiently.
Government should also take stand and be actively involved in the issues that are
related to the privacy and security to provide a clear and secure communication
[33]. Rana et al. [34] highlighted the need for transparency among citizens. For a
successful smart government, there are many key factors to be kept in mind such as
network integration, channel, smartphone services and smart city services [35]. We
must understand that government is very efficiently managing all the governmental
work and policies of residents, and this process helps the residents to be involved in
the evaluation of smart cities. There were various methods proposed, which calculate
the smartness of the smart cities by calculating the effectiveness, collaboration,
security, transparency and easiness [36]. For all the E-governance, an infrastructure
should be developed for better support over the cloud-based technology that will
enhance the data-processing time, decision-making based on evidence, sustainability,
resident’s centricities, creativity, equality, efficiency, openness, entrepreneurialism,
technology savviness, resiliency and innovation, which results in better and safe
smart governance [37].

17.2.4 Smart mobility

The potency of transport capacities based on the congestion in the smart cities is
one of the biggest tasks of the smart governance. Intelligent transportation man-
agement system is majorly discussed in the scientific articles given out by many
experts in the world [38]. Internet of vehicle (IoV) is the major solution that can be
considered at the solution of current problem and it also solves the issues like
efficiency and traffic safety. Zhu et al. [39] proposed a compatible network man-
agement system that the people of smart cities can easily adapt to the newly
emerging technology of IoV. It would be difficult to adapt to the IoV in the starting,
but this process grows much more efficient with time and by adaptation of people.
It was mentioned that urban traffic management solutions should be introduced
with a solution that drivers can reach to their destination without experiencing the
traffic problems [40]. Calderoni et al. [41] designed a model that will take the data
from different networks of the sensors by recognizing two different services,
namely smartphone-based traffic control system and the wise traffic controller.
These two systems do the wonderful job in smart cities with urban infrastructure, a
solution with the most efficient vehicle tracking management system in which data
can be collected on the basis of real-time movement of the vehicles in the smart
cities. However, improvisation in the model over time resulted in tracking of
multiple vehicles at a single time at much better accuracy [42].

17.2.5 Smart environment

Water supply, green spaces, waste management, air quality, monitoring, energy
efficiency, etc. all are elements of smart environment that need to be accomplished
in a way toward a better deployment of smart cities [43]. The major issue is the
336 The Internet of medical things

constant monitoring of the trees in the smart cities that will help one to build a
better smart environment. It is further noticed that trees in the smart cities could
also damage the network cables, and it also occurs disruption in power connections,
and some researchers suggested an efficient way, which is by developing a dynamic
laser-based scanning system, results in locating well-organized trees in the smart
cities [44]. Castelli et al. [45] discussed that the pollution in the air can cause severe
damage to the environment, which many urban cities are already facing. A report
published by World Health Organization says many people are dying because of air
pollution every year. However, making it very necessary to keep an eye on the air
quality index management, this could be done through the various traffic mon-
itoring sensors installed at the certain location in the smart cities [46]. The water
management is a challenging issue if not managed properly, as it has been seen in
the smart cities and few factors remain as a challenge such as maintenance cost,
new contaminants, and aging water infrastructure. Some experts suggest that ICT
systems improved the quality of drinking water across the smart cities. Water
management is directly responsible for the QoL that people are spending, as it is
also important for the governance to keep the contaminants as low as possible [47].

17.2.6 Smart living

Smart building for tourism, healthcare, public safety and education comes up
together as smart living. When we focus on these topics and work with better
governance, these sectors will automatically improve the QoL of the residents. The
other factor is the public safety of the people as it directly affects the development
of the smart cities. Interactive voice responses is a model that is used with the help
of citizens who can efficiently report the problem with the public safety, and it is
also very easy to learn [48]. Healthcare issues are very important to be assessed as
healthcare impacts the QoL of smart cities. Hussain et al. [49] proposed a frame-
work of hospital management that uses real-time monitoring for the disabled and
elderly people of the smart cities. Pramanik et al. [50] defined a demonstration for
the healthcare-connected business model that uses the trending technology like big
data for smart-healthcare system. This healthcare system helps the governance of
the smart cities to decrease the health costs, achieve better service and improve
contact management. Vincent [51] commented that the important factor, which is
education services, has a deep impact on the QoL of citizens of the smart cities at
the local development level. Another main factor is the smart buildings that are
very popular for their unique features and special design [52]. Table 17.1 presents
the components of smart cities that need to be achieved.

17.3 Computational infrastructures for smart cities big

data analytics
In few years, a great advancement is being noticed in especially DL techniques and
their accuracies. However, utmost memory and high computational power are required;
these things are majorly solved by some specific and advanced computational
Table 17.1 Components of smart cities

Smart people [22] Smart economy [28] Smart governance [33] Smart mobility Smart environment Smart living [48]
[38] [43]
● Healthcare study ● Employment ● IT connectivity ● Transport ● Pollution ● Sewerage and
● Telecommunication opportunities ● Online citizen services ● Walkability ● Green building sanitation
● Entertainment ● Smart manufacturing ● Digital media ● Traffic ● Renewable energy ● Water supply
● Cultural activities ● Smart data protection ● Wireless ● Smart grids ● Electricities
● Smart and advanced networks ● Air quality ● Building and home
security ● Trees ● Public open space
● Smart agriculture
sustainability
338 The Internet of medical things

platforms. Moreover, platforms like fog, edge and cloud computing are used. Edge
computing and fog computing are specially designed as extensions for the enabled data
analytics and cloud network for the source of the new data creation [52]. These three
computing techniques are explained in detail in the subsections.

17.3.1 Cloud computing

Cloud computing technology helps us to access the data from anywhere in the
world at any time; we can say the data and function at our fingertips. It contains
several data centers and servers that are available on the Internet for all the users in
the world [53]. There are many special features of IoT as discussed in the
following:
● Storage on the Internet: Cloud is a technology that connects the server to dif-
ferent devices through Transmission Control Protocol/Internet Protocol net-
works. Its major task is to help the movement of the data and to provide storage
deployment.
● Service over Internet: There are many services available on cloud technology,
majorly artificial intelligence (AI) and networking to support the different
needs of the users.
● Applications over Internet: Cloud applications help the user to perform tasks
through an Internet connection from anywhere instead of writing code, which
is not at all convenient.
The data created and generated by the devices are sent to cloud infrastructure,
where these data are stored and processed in real time by using various DL algo-
rithms [54]. Despite having many advantages such as sustainability, flexibility and
interoperability, cloud computing technology has few disadvantages as listed in the
following:
● Computation cost: The major two phases of DL models are inference and
training in which the data first is transferred to the cloud for these phases to
come into the effect. Both these processes use a high amount of bandwidth that
increases its cost significantly.
● Latency: To know the strength of the network, this can be said as the time
taken for the server to reply. Sometimes a cloud fails when it cannot respond in
a very short time. It causes a major problem with time-critical applications.
● Reliability: These services are provided wirelessly, which means any problem
at the network will affect the reliability of the services provided.

17.3.2 Fog computing

Fog computing is developed for different kinds of IoT applications and responsible
for data analytics and computing nodes formations. Vaquero et al. [55] discussed
that a large amount of heterogeneous decentralized devices are cooperated and
communicated with others universally that work with the network to perform
processing and storage tasks without any interruption from a third party. Fog
 Big data analytics and data mining 339

computing and cloud computing share the mutually similar service such as
deployment, networking and storage. Furthermore, fog computing technology is
only available in few geographical locations in the world, whereas cloud computing
is available everywhere without any complication of the geographical location.
Moreover, fog computing is specially designed for the use of real-time IoT appli-
cations. These types of computing help us with low latency, minimal issues with
privacy and security, and limited network bandwidth.

17.3.3 Edge computing

Edge computing is a new model that is developed to decrease and remove the cons
of cloud computing [56] as it is the extended version of the cloud computing. Edge
computing has a high computation space so that a huge amount of data are taken
and then processed at the edge before sending to the central server cloud. A diffi-
culty in the market is a huge increment in data creation that makes it a bit difficult
to transport the data to the cloud. To improve the data process, we perform ana-
lytics at the edge rather than sending it to the cloud. This series of steps is beneficial
in many ways such as it decreases the response time, improves bandwidth effi-
ciency, reduces energy utilization and also reduces network pressure [57]. Edge
computing is very important for a better development of the upcoming smart cities.
By using this technology, smart cities can improvise in the various sectors and
people can use edge gateway by which all the smart device stands connected to the
network. All the data are meant to be processed at local level, which will increase
the security, privacy and overall performances.

17.3.4 Big data based on machine learning

Big data is available in all places and sectors around the world. Conversely, its
difficulty exceeds the processing power of outdated tools. For this reason, high-
performance computing platforms are mandatory to use the full capacities of big
data. These needs have unquestionably converted into a real challenge. Numerous
revisions focus on searching a method that permits reducing of computational costs
by increasing the compliance of the extracted data [58]. The necessity to excerpt
useful information has required using of different methods of ML, to compare the
results obtained and to analyze them according to the features of huge data
volumes. ML can perform any task that a human is capable of doing with his
human intelligence. Major of this intelligence revolves around few important fac-
tors such as engrossing speech to text, translating any language and image recog-
nition. The power of AI has grown significantly in the last few years [59]. AI
provides the capability to the computer to enact and take the decision through
computer programs by improving the learning behavioral changes after trained
unlabeled datasets from the mixed programs. At a later stage, the finally trained
program computer now has the ability to take new decisions. ML models can learn
nonlinear functions and features from a huge amount of complex, raw data with
much better performance than over traditional ML methods. ML can be broadly
classified into four major categories, discussed as follows:
340 The Internet of medical things

Table 17.2 Categories of ML

Supervised Unsupervised Semi-supervised Reinforcement

learning [60] learning [60] learning [61] learning [61]

● Known labels ● Unknown labels ● Known labels or ● Focus on making

or output or output output for a decision, based
● Focus on finding subset of data on previous
patterns and ● A blend of experience
gaining insight supervised and ● Policy-making
from the data unsupervised with feedback
learning
Uses: Uses: Uses: Uses:
● Insurance ● Customer ● Medical ● Gaming
underwriting clustering prediction ● Complex
● Fraud detection ● Association decision making
rule mining ● Reward system

● Supervised learning: It refers to the capability of learning from labeled training

data and helps in the way of predicting the outcomes majorly for unforeseen data.
● Unsupervised learning: It refers to the capability of examining the relationship
and the naturally descriptive statistic of the data. The unsupervised learning is
utilized to find out the interesting, hidden structure of the data [60].
● Semi-supervised learning: It refers to the training of a huge amount of data,
which mainly contain less labeled data as compared to fully unlabeled data.
● Reinforcement learning: It occurs in between unsupervised and supervised
learning and gives the ability to an “agent” to extract the learning from its
nearby surrounding and the environment, which in turn makes the agent highly
capable of behaving and acting smartly, then the “agent” can provide direct
interaction with environment by its “actions” and accepts “reward” from
nearby surrounding as well as “penalty” for unwanted results to create and
inspire from a very perfect behavior policy [61]. Table 17.2 presents the
summarized view of above discussed ML categories.

17.4 Mining methods for big data

There are different kinds of heterogeneous data available in our surroundings and
we need to apply the DM techniques to utilize all these heterogeneous available
data, with the help of automation [62]. Some mining techniques that come under
the supervised, unsupervised and reinforcement learning are discussed next.

17.4.1 Classification
In classification, a process is followed through which an object is assigned to the
predefined category, classification techniques can predict the destination class of
 Big data analytics and data mining 341

each object of the data. For supervised learning, the targeted labels are assumed
before processing. Before the classification of unknown data/objects or unlabeled
data, a classifier (prediction function) classifies the required training as most
algorithms are classified into two types. The first one is computing the probability
of an item whether it belongs to a particular class or not and the second is com-
paring it with the cut-off value [63]. The evaluation of the performance of classi-
fication model can be done as follows:


number of correct prediction Tp þ Fp
Accuracy ¼
(17.1)
total number of predictions Tp þ Fp þ Tn þ Fn

number of wrong prediction ðTn þ Fn Þ

Error rate ¼
(17.2)
total number of predictions Tp þ Fp þ Tn þ Fn
Precision is an alternative method through which we can find out the classi-
fication algorithm results. Precision can be defined as the probability that selection
taken on a random basis always results in relevance and recall can be defined as if
the probability of a randomly selected object is reclaimed [64]. The mathematical
expression for both is as follows:
Tp
Precision ðprÞ ¼ (17.3)
Tp þ Fp
Tp
Recall ðRÞ ¼ (17.4)
Tp þ Fn
Here, Tn , Tp , Fn and Fp can be truly said as a confusion matrix as shown in
Table 17.3.
Based on both recall and precision, these classification results can be given an
F-score by the following formula:
2  pr  R
F  score ¼ (17.5)
pr þ R

17.4.2 Clustering
A cluster can be defined as the objects present in a special identical group. With the
help of clustering algorithm, the clusters are segregated into a certain fixed number
of subclusters on the basis of their features. Clustering follows unsupervised

Table 17.3 Confusion matrix

Prediction class ¼ 0 Prediction class ¼ 0

Actual class ¼ 1 False negative (Tn) True positive (Tp)
Actual class ¼ 0 True negative (Tn) False positive (Fp)
342 The Internet of medical things

learning technique, for example, in few medical centers, certain patients have an
unknown disease but researchers of the medical field are able to uncover and
provide cure of those diseases as they are having little data related to the symptom
of disease and improvement made by the patients based on some treatments that are
provided by the doctors [65]. This clustering will help in a way to segregate the
patients on the basis of their recognized symptoms and progress, which will help
the doctors to perform a better and improved treatment.

17.4.3 Frequent pattern mining

Frequent pattern mining can be defined as a sequence of events or a set of data
objects or data object that appears frequently on the system is known as a frequent
pattern. When mining is done on frequent patterns, it gives a good overview of the
analytical section and understanding of the data insights of the citizen’s activity in a
much suitable environment. Analyzing the pattern always plays an important role
by which we can correlate and find the relationship between the data [66]. With the
broad application of the association rule, one can easily predict that how the cus-
tomer purchases a new item. This can yield out a complete prediction of the cus-
tomer’s pattern of purchase, which will help one to ramp up the customer
experience and boost the business sales. The patterns with some specific orders can
be called sequential patterns and when we perform mining on these patterns then it
is called sequential pattern mining and can be applied to transaction patterns of any
type of customer [67].

17.4.4 Another mining method

As outliers present in the data, during DM, lots of exceptions have to be faced.
Sometimes these outliers can also be called anomalous objects. Outliers always have
a different identity while comparing it with the regular data, and due to this unique
identity, it is called an outlier [68]. By looking at the properties, especially the
inherent properties, it can find out the good insights over the data. For outlier
detection, data can take from IoT applications like smart traffic, parking system,
smart agriculture, healthcare and smart home. Broadly outlier detection can be seen
through four methods: distance-based, statistical-based, distribution, deviation and
density local outlier detection. Outlier mining method is used in E-health monitoring
system, in which health parameter is calculated via metric sensors and analyzed. All
these data are collected in one place and then mining is done for outlier detection. It
helps one to extract all the anomalous information present in the data, which will help
the healthcare management system to predict all emergencies [69].

17.5 Advances in healthcare sector

Data in the healthcare sector is unstructured as these come from various healthcare
applications such as health information exchange, public records, genetic data-
bases, EHR and sensor data. Owing to the huge volume, heterogeneous data
 Big data analytics and data mining 343

formats, and other associated uncertainties, it becomes difficult to transform the

raw data into proper information. Selection of class attributes and health feature
identification in healthcare data requires highly architecturally specific and
sophisticated tools and techniques. The applicability of BDA and DM in healthcare
sector is explored in the following subsections.

17.5.1 Big data for healthcare

Owing to the growing global population and the consequent generation of volu-
minous data in a healthcare sector, big data has emerged as a new promising
solution. Pharmaceutical-industry stakeholders and experts have actively started
analyzing big data to improve healthcare quality. Big data-enabled healthcare
systems have been established across numerous disciplines such as treatment cost
determination and patient characteristic investigation. Health informatics involves
the storage, retrieval and acquisition of data in order to yield efficient results to
service providers. Healthcare data are heterogeneous and link to numerous bio-
medical sources, including demographics, laboratory tests, gene arrays, imagery
and sensors. Mostly these data are unstructured and unfit for storing electronically.
This leads to the concept of data digitization that is supported by the four Vs,
namely veracities, velocities, volume and variety [70,71].
Viceconti et al. [71] combined BDA and virtual psychological human tech-
nologies to produce effective and robust medical solutions that are capable of
working with heterogeneous data spaces, non-textual information and sensitive
data. In another work, Zhang et al. [72] proposed a cyber-physical system–based
healthcare application built on BDA and cloud computing. Similarly, Hossain and
Muhammad [73] proposed a powerful emotion detection module for connected
healthcare system that can capture image and speech signals of patients within
smart homes. Sahoo et al. [74] proposed an improved naı̈ve Bayes algorithm that
uses cloud big data for the prediction of future health conditions. The work also
proposes an optimal data distribution scheme for positioning both streaming and
batch patient data toward spark nodes. In another work, Hussain et al. [75] pro-
posed a novel scheme to avoid semantic loss during healthcare document parti-
tioning in big data. The proposed model set up uniquely configured systems using
clinical document architecture on HDFS. Similarly, Syed et al. [76] proposed a
novel healthcare framework to help elderly people in ambient-assisted living
scenarios. The proposed framework achieved effective decision-making, analysis,
treatment recommendations. In another work, Zhou et al. [77] proposed a fre-
quent pattern mining-based medical cluster behavior model by combining the
association rule mining with MapReduce computing model. Nazir et al. [78]
summarized the existing research work and discussed the issues related to large-
scale data management, complex system modeling, processing voluminous data
and derivations from healthcare simulations and data. Kim and Chung [79] pro-
posed to apply a multimodal autoencoder in healthcare big data for missing data
estimation. The work used denoising an autoencoder to find missing values
in PHR.
Table 17.4 Big data analytics and data mining for healthcare sector

Technologies Reference Year Contribution

Big data for healthcare Viceconti et al. [71] 2015 Robust medical solutions that are capable of working with
heterogeneous data spaces, non-textual information and sensi-
tive data
Zhang et al. [72] 2017 Healthcare applications built on big data analytics and cloud
computing
Hossain and Muhammad [73] 2018 Emotion detection module for connected healthcare system
Sahoo et al. [74] 2018 Cloud big data for prediction of future health conditions
Hussain et al. [75] 2019 Avoid semantic loss during healthcare document partitioning in
big data
Syed et al. [76] 2019 Healthcare framework to help elderly people in ambient-assisted
living scenarios
Zhou et al. [77] 2020 Medical cluster behavior model
Nazir et al. [78] 2020 Discussed the issues related to large-scale data management
Kim and Chung [79] 2020 Multimodal autoencoder in healthcare big data for missing data
estimation
Data mining for healthcare Akay et al. [81] 2015 Framework for cancer treatment using network partitioning
scheme
Yassine et al. [82] 2017 Big data–based smart home model for measuring and analyzing
energy used in occupants behavior
Kovalchuk et al. [83] 2018 Hybrid simulation–based framework for the discovery and
classification of clinical pathways for patients
Amin et al. [84] 2019 Hybrid data mining–based prediction model for predicting heart
disease
Sun et al. [85] 2019 Network-based progressing mining method for chronic diseases
Gonçalves et al. [86] 2020 Predict mental illness considering unemployment factor
Kaur et al. [87] 2020 Prostate cancer prediction model
Zhang et al. [88] 2020 Enhanced privacy data fusion scheme that employs mining
solutions
 Big data analytics and data mining 345

17.5.2 Data mining for healthcare

The DM methods can be helpful for pattern discovery in order to generate com-
putational tools and theories that assist humans in useful knowledge extraction
from the ever-growing voluminous data. The most common DM methods that are
used in healthcare sector are data visualization, pattern matching, evolution, asso-
ciation, clustering, generalization, classification, meta-rule-guided mining and
characterization. Considering the underutilization of healthcare data, the concepts
of DM can be used to extract valuable information and thereby effectively predict
and diagnose disease [80].
Akay et al. [81] proposed a dual-analysis framework for cancer treatment that
focuses on treatment’s side effects and forum posts. Word frequency data in forum
posts are analyzed using the self-organizing maps, and stability measurement is
achieved using network partitioning scheme. In another work, Yassine et al. [82]
proposed a big data–based smart home model that facilitates easy discovering and
learning of human activity patterns. The proposed scheme utilized prediction,
cluster analysis and frequent pattern mining for measuring and analyzing energy
used in occupants behavior. Kovalchuk et al. [83] proposed a hybrid simulation-
based framework for the discovery and classification of clinical pathways for
patients. Amin et al. [84] proposed a hybrid DM-based prediction model for pre-
dicting heart disease using seven classification algorithms and DM techniques. In
another work, Sun et al. [85] proposed a network-based progressing mining method
to improve the clarity on chronic diseases and the associated medication stages.
Gonçalves et al. [86] aimed to predict mental illness and associate it with factors
like unemployment. The work integrates the concept of DM and cross-industry
standard protocol to identify the factors associated with mental illness prediction.
Similarly, Kaur et al. [87] used Indian datasets to develop a prostate cancer pre-
diction model and highlight various predictors related to cancer survivability.
Zhang et al. [88] proposed an enhanced privacy data fusion scheme that employs
mining solutions and multisource data integration to provide better and timely
health services to patients. Table 17.4 summarizes various contributions that
highlight the role of big data and DM in the healthcare sector.

17.6 Conclusion

Technologies like big data, ML, AI, DL and DM create a deep impact on people’s
lifestyle. The development of the smart cities is highly proportional to such tech-
nologies. All these technologies can be clubbed together with the IoT devices to
provide more efficiency. Many issues present in the real-world cities can be solved
with these technologies in the new upcoming smart cities. For example, the issues
like traffic management can be resolved with the fully automated robot system and
by designing it with a better infrastructure, which will increase the efficiency in
traffic management. Moreover, sustainable concepts of smart cities should carry
keywords like environment, land use, transportation and relations with each other.
The designer of the smart cities should have few aspects in his mind before
346 The Internet of medical things

designing the smart cities’ layout such as the population of the smart cities, size of
the smart cities, how much it is accessible to the service center of the prominent
company and how much accessible it is to the other cities that exist nearby. The
main aim of our study through this chapter was to spread the awareness among the
people that what are the key future trends of the smart cities and how we can build a
smart healthcare to provide better treatment with ease. The chapter has described a
theoretical background of smart cities in which some important components are
discussed. Furthermore, the chapter highlights the need of computational infra-
structure for smart cities in a correlation with BDA. Moreover, the chapter high-
lights some mining methods for mining of big as well as heterogeneous data.
Additionally, the chapter has described the advancement of healthcare sector in
smart cities in context of BD and DM techniques. In future, we enhance this study
in a way of providing more security and ease.

References

[1] Agbali M., Trillo C., Fernando T., Oyedele L., Ibrahim I. A., and Olatunji V. O.
(2019). Towards a refined conceptual framework model for a smart and sus-
tainable cities assessment. IEEE International Smart Cities Conference (ISC2),
pp. 658–664, doi:10.1109/ISC246665.2019.9071697.
[2] Suryanegara M., Prasetyo D. A., Andriyanto F., and Hayati N. (2019). A
5-step framework for measuring the quality of experience (QoE) of Internet
of Things (IoT) services. IEEE Access, 7, 175779–175792, doi:10.1109/
ACCESS.2019.2957341.
[3] Mishra K. N. and Chakraborty C. (2020). A novel approach towards using
big data and IoT for improving the efficiency of m-health systems. In: Gupta
D., Hassanien A., Khanna A. (eds), Advanced Computational Intelligence
Techniques for Virtual Reality in Healthcare. Studies in Computational
Intelligence, vol. 875. Cham: Springer. https://doi.org/10.1007/978-3-030-
35252-3_7.
[4] Gupta A., Chakraborty C., and Gupta B. (2019). Medical information pro-
cessing using smartphone under IoT framework. In: Mittal M., Tanwar S.,
Agarwal B., Goyal L. (eds), Energy Conservation for IoT Devices. Studies in
Systems, Decision and Control, vol. 206. Singapore: Springer. https://doi.
org/10.1007/978-981-13-7399-2_12.
[5] Mishra K. N. and Chakraborty C. (2020). A novel approach toward enhan-
cing the quality of life in smart cities using clouds and IoT-based technolo-
gies. In: Farsi M., Daneshkhah A., Hosseinian-Far A., Jahankhani H. (eds),
Digital Twin Technologies and Smart Cities. Internet of Things (Technology,
Communications and Computing). Cham: Springer. https://doi.org/10.1007/
978-3-030-18732-3_2.
[6] Gupta A., Chakraborty C., and Gupta B. (2019). Monitoring of epileptical
patients using cloud-enabled health-IoT system. Traitement Du Signal, 36(5),
425–431. doi:10.18280/ts.360507.
 Big data analytics and data mining 347

[7] Chakraborty C. and Gupta B. (2016). Adaptive filtering technique for

chronic Wound analysis UNDER Tele-Wound Network. Journal of
Communication, Navigation, Sensing and Services (CONASENSE), (1), 57–
76. doi:10.13052/jconasense2246-2120.2016.005.
[8] Haque A. K., Bhushan B., and Dhiman G. (2021). Conceptualizing smart
city applications: Requirements, architecture, security issues, and emerging
trends. Expert Systems. https://doi.org/10.1111/exsy.12753.
[9] Huynh-The T., Hua C., Doan V., Pham Q., Nguyen T., and Kim D. (2020). Deep
learning for Constellation-based modulation classification under multipath fading
channels. 2020 International Conference on Information and Communication
Technology Convergence (ICTC). doi:10.1109/ictc49870.2020.9289413.
[10] Vinayakumar R., Alazab M., Srinivasan S., Pham Q., Padannayil S. K., and
Simran K. (2020). A visualized botnet detection system based deep learning
for the internet of things networks of smart cities. IEEE Transactions on
Industry Applications, 56(4), 4436–4456. doi:10.1109/tia.2020.2971952.
[11] Alazab M., Khan S., Krishnan S. S., Pham Q., Reddy M. P., and Gadekallu
T. R. (2020). A multidirectional LSTM model for predicting the stability
of a smart grid. IEEE Access, 8, 85454–85463. doi:10.1109/access.2020.
2991067.
[12] Nduwayezu M., Pham Q., and Hwang W. (2020). Online computation off-
loading in NOMA-based multi-access edge computing: A deep reinforce-
ment learning approach. IEEE Access, 8, 99098–99109. doi:10.1109/
access.2020.2997925.
[13] Goyal S., Sharma N., Bhushan B., Shankar A., and Sagayam M. (2021). IoT
enabled technology in secured healthcare: Applications, challenges and
future directions. In: Hassanien A.E., Khamparia A., Gupta D., Shankar K.,
Slowik A. (eds), Cognitive Internet of Medical Things for Smart Healthcare.
Studies in Systems, Decision and Control, vol. 311. Cham: Springer. https://
doi.org/10.1007/978-3-030-55833-8_2.
[14] Goyal S., Sharma N., Kaushik I., Bhushan B., and Kumar A. (2020).
Precedence & Issues of IoT based on Edge Computing. 2020 IEEE 9th
International Conference on Communication Systems and Network
Technologies (CSNT). doi:10.1109/csnt48778.2020.9115789.
[15] Sharma N., Kaushik I., Bhushan B., Gautam S., and Khamparia A. (2020).
Applicability of WSN and biometric models in the field of healthcare. Deep
Learning Strategies for Security Enhancement in Wireless Sensor Networks
Advances in Information Security, Privacy, and Ethics, IGI Global, USA,
304–329. doi:10.4018/ 978-1-7998-5068-7.ch016.
[16] Bhushan B., Sahoo C., Sinha P., and Khamparia A. (2020). Unification of
Blockchain and Internet of Things (BIoT): Requirements, working model, chal-
lenges and future directions. Wireless Networks. doi:10.1007/s11276-020-02445-6.
[17] Saxena S., Bhushan B., and Ahad M. A. (2021). Blockchain based solutions
to Secure IoT: Background, integration trends and a way forward. Journal
of Network and Computer Applications, 103050. doi:10.1016/j.jnca.2021.
103050.
348 The Internet of medical things

[18] Bhushan B., Khamparia A., Sagayam K. M., Sharma S. K., Ahad M. A., and
Debnath N. C. (2020). Blockchain for smart cities: A review of architectures,
integration trends and future research directions. Sustainable Cities and
Society, 61, 102360. doi:10.1016/j.scs.2020.102360.
[19] Khamparia A., Singh P. K., Rani P., Samanta D., Khanna A., and Bhushan B.
(2020). An internet of health things-driven deep learning framework for
detection and classification of skin cancer using transfer learning.
Transactions on Emerging Telecommunications Technologies. doi:10.1002/
ett.3963.
[20] Lazaroiu G. C. and Roscia M. (2012). Definition methodology for the smart
cities model. Energy, 47(1), 326–332. doi:10.1016/j.energy.2012.09.028.
[21] Sethi R., Bhushan B., Sharma N., Kumar R., and Kaushik I. (2021).
Applicability of industrial IoT in diversified sectors: Evolution, applications
and challenges. In: Kumar R., Sharma R., Pattnaik P.K. (eds), Multimedia
Technologies in the Internet of Things Environment, Studies in Big Data,
vol. 79. SIngapore: Springer. https://doi.org/10.1007/978-981-15-7965-3_4.
[22] Malik A., Gautam S., Abidin S., and Bhushan B. (2019). Blockchain
technology-future of IoT: Including structure, limitations and various pos-
sible attacks. 2nd International Conference on Intelligent Computing,
Instrumentation and Control Technologies (ICICICT), Kannur, India, 2019,
pp. 1100–1104, doi:10.1109/ICICICT46008.2019.8993144.
[23] Yeh H. (2017). The effects of successful ICT-based smart cities services:
From citizens’ perspectives. Government Information Quarterly, 34(3), 556–
565. https://doi.org/10.1016/j.giq.2017.05.001.
[24] Belanche-Gracia D., Casaló-Ariño L. V., and Pérez-Rueda A. (2015).
Determinants of multi-service smartcard success for smart cities develop-
ment: A study based on citizens’ privacy and security perceptions.
Government Information Quarterly, 32(2), 154–163. https://doi.org/10.1016/
j.giq.2014.12.004.
[25] Chatterjee S. and Kar A. K. (2017). Effects of successful adoption of
information technology enabled services in proposed smart cities of India.
Journal of Science and Technology Policy Management, 9(2), 189–209.
https://doi.org/10.1108/jstpm-03-2017-0008.
[26] Niforatos E., Vourvopoulos A., and Langheinrich M. (2017). Understanding
the potential of human–machine crowdsourcing for weather data.
International Journal of Human-Computer Studies, 102, 54–68. https://doi.
org/10.1016/j.ijhcs.2016.10.002.
[27] Mone G. (2015). The new smart cities. Communications of the ACM, 58(7),
20–21. https://doi.org/10.1145/2771297.
[28] Gautam S., Malik A., Singh N., and Kumar S. (2019). Recent advances and
countermeasures against various attacks in IoT environment. 2019 2nd
International Conference on Signal Processing and Communication
(ICSPC). https://doi.org/10.1109/icspc46172.2019.8976527.
[29] Johnson P., Iacob M. E., Välja M., van Sinderen M., Magnusson C., and
Ladhe T. (2014). A method for predicting the probability of business
 Big data analytics and data mining 349

network profitability. Information Systems and e-Business Management, 12

(4), 567–593. https://doi.org/10.1007/s10257-014-0237-4.
[30] Keegan S., O’Hare G. M. P., and O’Grady M. J. (2012). Retail in the digital
cities. International Journal of E-Business Research, 8(3), 18–32. https://
doi.org/10.4018/jebr.2012070102.
[31] An J., Le Gall F., Kim J., Yun J., Hwang J., Bauer M., and Song J. (2019).
Toward global IoT-enabled smart cities interworking using adaptive
semantic adapter. IEEE Internet of Things Journal, 6(3), 5753–5765. https://
doi.org/10.1109/jiot.2019.2905275.
[32] Dı́az-Dı́az R. and Pérez-González D. (2019). Implementation of social
media concepts for E-government: Case study of a social media tool for
value co-creation and citizen participation. In I. Management Association
(Ed.), Smart Cities and Smart Spaces: Concepts, Methodologies, Tools, and
Applications (pp. 1071–1091). IGI Global. http://doi:10.4018/978-1-5225-
7030-1.ch049.
[33] Gascó-Hernandez M. (2018). Building a smart cities. Communications of the
ACM, 61(4), 50–57. https://doi.org/10.1145/3117800.
[34] Rana N. P., Luthra S., Mangla S. K., Islam R., Roderick S., and Dwivedi Y. K.
(2018). Barriers to the development of smart cities in Indian context. Information
Systems Frontiers, 21(3), 503–525. https://doi.org/10.1007/s10796-018-9873-4.
[35] Cledou G., Estevez E., and Soares Barbosa L. (2018). A taxonomy for
planning and designing smart mobility services. Government Information
Quarterly, 35(1), 61–76. https://doi.org/10.1016/j.giq.2017.11.008.
[36] Fietkiewicz K. J., Mainka A., and Stock W. G. (2017). eGovernment in cities
of the knowledge society. An empirical investigation of Smart Cities’ gov-
ernmental websites. Government Information Quarterly, 34(1), 75–83.
https://doi.org/10.1016/j.giq.2016.08.003.
[37] Guetat S. B. and Dakhli S. B. (2016). Services-based integration of urbanized
information systems. Information Resources Management Journal, 29(4),
17–34. https://doi.org/10.4018/irmj.2016100102.
[38] Sussman F. S. (2001). Telecommunications and transnationalism: The
polarization of social space. The Information Society, 17(1), 49–62. https://
doi.org/10.1080/019722401750067423.
[39] Zhu W., Gao D., Zhao W., Zhang H., and Chiang H.-P. (2017). SDN-enabled
hybrid emergency message transmission architecture in internet-of-vehicles.
Enterprise Information Systems, 12(4), 471–491. https://doi.org/10.1080/
17517575.2017.1304578.
[40] Walravens N. (2012). Mobile business and the smart cities: Developing a
business model framework to include public design parameters for mobile
cities services. Journal of Theoretical and Applied Electronic Commerce
Research, 7(3), 23–24. https://doi.org/10.4067/s0718-18762012000300012.
[41] Calderoni L., Maio D., and Rovis S. (2014). Deploying a network of
smart cameras for traffic monitoring on a “cities kernel”. Expert Systems
with Applications, 41(2), 502–507. https://doi.org/10.1016/j.eswa.2013.
07.076.
350 The Internet of medical things

[42] Lee G., Mallipeddi R., and Lee M. (2017). Trajectory-based vehicle tracking
at low frame rates. Expert Systems with Applications, 80, 46–57. https://doi.
org/10.1016/j.eswa.2017.03.023.
[43] Anagnostopoulos T., Kolomvatsos K., Anagnostopoulos C., Zaslavsky A.,
and Hadjiefthymiades S. (2015). Assessing dynamic models for high priority
waste collection in smart cities. Journal of Systems and Software, 110, 178–
192. https://doi.org/10.1016/j.jss.2015.08.049.
[44] Corbett J. and Mellouli S. (2017). Winning the SDG battle in cities: How an
integrated information ecosystem can contribute to the achievement of the
2030 sustainable development goals. Information Systems Journal, 27(4),
427–461. https://doi.org/10.1111/isj.12138.
[45] Castelli M., Gonçalves I., Trujillo L., and Popovic A. (2016). An evolu-
tionary system for ozone concentration forecasting. Information Systems
Frontiers, 19(5), 1123–1132. https://doi.org/10.1007/s10796-016-9706-2.
[46] Polenghi-Gross I., Sabol S. A., Ritchie S. R., and Norton M. R. (2014).
Water storage and gravity for urban sustainability and climate readiness.
Journal – American Water Works Association, 106(12). https://doi.org/10.
5942/jawwa.2014.106.0151.
[47] Breetzke T. and Flowerday S. V. (2016). The usability of IVRs for smart
cities crowdsourcing in developing cities. The Electronic Journal of
Information Systems in Developing Countries, 73(1), 1–14. https://doi.org/
10.1002/j.1681-4835.2016.tb00527.x.
[48] Cilliers L. and Flowerday S. (2017). Factors that influence the usability of a
participatory IVR crowdsourcing system in a smart cities. South African
Computer Journal, 29(3). https://doi.org/10.18489/sacj.v29i3.422.
[49] Hussain A., Wenbi R., da Silva A. L., Nadher M., and Mudhish M. (2015).
Health and emergency-care platform for the elderly and disabled people in
the smart cities. Journal of Systems and Software, 110, 253–263. https://doi.
org/10.1016/j.jss.2015.08.041.
[50] Pramanik M. I., Lau, R. Y. K., Demirkan H., and Azad M. A. (2017). Smart
health: Big data enabled health paradigm within smart cities. Expert Systems
with Applications, 87, 370–383. https://doi.org/10.1016/j.eswa.2017.06.027.
[51] Vincent J. M. (2006). Public schools as public infrastructure. Journal of
Planning Education and Research, 25(4), 433–437. https://doi.org/10.1177/
0739456x06288092.
[52] Mukherjee M., Shu L., and Wang D. (2018). Survey of fog computing:
Fundamental, network applications, and research challenges. IEEE
Communications Surveys & Tutorials, 20(3), 1826–1857. https://doi.org/10.
1109/comst.2018.2814571.
[53] Bonomi F., Milito R., Zhu J., and Addepalli, S. (2012). Fog computing and
its role in the internet of things. Proceedings of the First Edition of the MCC
Workshop on Mobile Cloud Computing – MCC ‘12. https://doi.org/10.1145/
2342509.2342513.
[54] Rashid M. and Wani U. I. (2020). Role of fog computing platform in ana-
lytics of internet of things- issues, challenges and opportunities. Fog, Edge,
 Big data analytics and data mining 351

and Pervasive Computing in Intelligent IoT Driven Applications, pp. 209–220.

Wiley. https://doi.org/10.1002/9781119670087.ch12.
[55] Vaquero L. M. and Rodero-Merino L. (2014). Finding your way in the fog.
ACM SIGCOMM Computer Communication Review, 44(5), 27–32. https://
doi.org/10.1145/2677046.2677052.
[56] Yu W., Liang F., He X., Hatcher W. G., Lu C., Lin J., and Yang X. (2018). A
survey on the edge computing for the internet of things. IEEE Access, 6,
6900–6919. https://doi.org/10.1109/access.2017.2778504.
[57] Wang X., Han Y., Leung V. C., Niyato D., Yan X., and Chen X. (2020).
Convergence of edge computing and deep learning: A comprehensive sur-
vey. IEEE Communications Surveys & Tutorials, 22(2), 869–904. https://
doi.org/10.1109/comst.2020.2970550.
[58] Qiu J., Wu Q., Ding G., Xu Y., and Feng S. (2016). Erratum to: A survey
of machine learning for big data processing. EURASIP Journal on
Advances in Signal Processing, 2016(1). https://doi.org/10.1186/s13634-
016-0382-7.
[59] Garcia-Almanza A. L., Alexandrova-Kabadjova B., and Martinez-Jaramillo S.
(2013). Bankruptcy Prediction for Banks: An Artificial Intelligence Approach
to Improve Understandability. Studies in Computational Intelligence, 633–656.
https://doi.org/10.1007/978-3-642-29694-9_24.
[60] Wen Chen X. and Lin X. (2014). Big data deep learning: Challenges and
perspectives. IEEE Access, 2, 514–525. https://doi.org/10.1109/access.2014.
2325029.
[61] Shokri R. and Shmatikov V. (2015). Privacy-preserving deep learning. 2015
53rd Annual Allerton Conference on Communication, Control, and
Computing (Allerton). https://doi.org/10.1109/allerton.2015.7447103.
[62] Brynjolfsson E. and Mcafee A. (2017). The business of artificial intelligence,
Harvard Bus. Rev. 2017 Artificial Intelligence and Robotics (IRANOPEN).
https://doi.org/10.1109/rios.2017.7956467.
[63] James G., Witten D. Hastie T., and Tibshirani R. (2013). Unsupervised
Learning. Springer Texts in Statistics, 373–418. https://doi.org/10.1007/978-
1-4614-7138-7_10.
[64] Hastie T., Tibshirani R., and Friedman J. (2008). Unsupervised Learning.
The Elements of Statistical Learning, 485–585. https://doi.org/10.1007/978-
0-387-84858-7_14.
[65] Kingma P., Rezende D. J., Mohamed D. S., and Welling M. (n.d.). Semi-
Supervised Learning with Deep Generative Models. https://doi.org/
arXiv:1406.5298.
[66] Van Hasselt H., Guez A., and Silver D. (2016). Deep reinforcement learning
with double Q-learning. Proceedings of the AAAI Conference on Artificial
Intelligence, 30(1). Retrieved from https://ojs.aaai.org/index.php/AAAI/
article/view/10295.
[67] Bassolas A., Barbosa-Filho H., Dickinson B., et al. (2019). Hierarchical
organization of urban mobility and its connection with cities livability. Nature
Communications, 10(1). https://doi.org/10.1038/s41467-019-12809-y.
352 The Internet of medical things

[68] Li R., Zhao Z., Yang C., Wu C., and Zhang H. (2018). Wireless big data in
cellular networks: The cornerstone of smart cities. IET Communications, 12
(13), 1517–1523. https://doi.org/10.1049/iet-com.2017.1278.
[69] Luo F., Cao G., Mulligan K., and Li X. (2016). Explore spatiotemporal and
demographic characteristics of human mobility via Twitter: A case study of
Chicago. Applied Geography, 70, 11–25. https://doi.org/10.1016/j.apgeog.
2016.03.001.
[70] O’Sullivan F. and Holzinger S. A. (2021). AI and big data in health-
care: Towards a more comprehensive research framework for multi-
morbidity. Journal of Clinical Medicine, 10(4), 766. doi:10.3390/
jcm10040766.
[71] Viceconti M., Hunter P., and Hose R. (2015). Big data, big knowledge: Big
data for personalized healthcare. IEEE Journal of Biomedical and Health
Informatics, 19(4), 1209–1215. doi:10.1109/JBHI.2015.2406883.
[72] Zhang Y., Qiu M., Tsai C. W., Hassan M. M., and Alamri A. (2017).
Health-CPS: Healthcare cyber-physical system assisted by cloud and big
data. IEEE Systems Journal, 11(1), 88–95. doi:10.1109/JSYST.2015.
2460747.
[73] Hossain M. S. and Muhammad G. (2018). Emotion-aware connected
healthcare big data towards 5G. IEEE Internet of Things Journal, 5(4),
2399–2406. doi:10.1109/JIOT.2017.2772959.
[74] Sahoo P. K., Mohapatra S. K., and Wu S. (2018). SLA based
HEALTHCARE big data analysis and computing in cloud network. Journal
of Parallel and Distributed Computing, 119, 121–135. doi:10.1016/j.
jpdc.2018.04.006.
[75] Hussain S., Hussain M., Afzal M., Hussain J., Bang J., Seung H., and Lee S.
(2019). Semantic preservation of standardized healthcare documents in big
data. International Journal of Medical Informatics, 129, 133–145.
doi:10.1016/j.ijmedinf.2019.05.024.
[76] Syed L., Jabeen S. S. M., and Alsaeedi A. (2019). Smart healthcare frame-
work for AMBIENT assisted living using IoMT and big data analytics
techniques. Future Generation Computer Systems, 101, 136–151.
doi:10.1016/j.future.2019.06.004.
[77] Zhou S., He J., Yang H., Chen D., and Zhang R. (2020). Big data-driven
abnormal behavior detection in healthcare based on association rules. IEEE
Access, 8, 129002–129011. doi: 10.1109/ACCESS.2020.3009006.
[78] Nazir S., Khan S., Khan H. U., et al. (2020). A comprehensive analysis of
healthcare big data management, analytics and scientific programming.
IEEE Access, 8, 95714–95733. doi:10.1109/access.2020.2995572.
[79] Kim J. and Chung K. (2020). Multi-modal stacked denoising autoencoder for
handling missing data in healthcare big data. IEEE Access, 8, 104933–
104943. doi: 10.1109/ACCESS.2020.2997255.
[80] Jothi N., Rashid N. A., and Husain W. (2015). Data mining in healthcare – A
review. Procedia Computer Science, 72, 306–313. doi:10.1016/j.procs.
2015.12.145.
 Big data analytics and data mining 353

[81] Akay A., Dragomir A., and Erlandsson B. (2015). Network-based modeling
and intelligent data mining of social media for improving care. IEEE Journal
of Biomedical and Health Informatics, 19(1), 210–218. doi:10.1109/
JBHI.2014.2336251.
[82] Yassine A., Singh S., and Alamri A. (2017). Mining human activity patterns
from smart home big data for health care applications. IEEE Access, 5,
13131–13141. doi: 10.1109/ACCESS.2017.2719921.
[83] Kovalchuk S. V., Funkner A. A., Metsker O. G., and Yakovlev A. N. (2018).
Simulation of patient flow in multiple healthcare units using process and
data mining techniques for model identification. Journal of Biomedical
Informatics, 82, 128–142. doi:10.1016/j.jbi.2018.05.004.
[84] Amin M. S., Chiam Y. K., and Varathan K. D. (2019). Identification of
significant features and data mining techniques in predicting heart disease.
Telematics and Informatics, 36, 82–93. doi:10.1016/j.tele.2018.11.007.
[85] Sun C., Li Q., Cui L., Li H., and Shi Y. (2019). Heterogeneous network-
based chronic disease progression mining. Big Data Mining and Analytics,
2(1), 25–34. doi:10.26599/BDMA.2018.9020009.
[86] Gonçalves C., Ferreira D., Neto C., Abelha A., and Machado J. (2020).
Prediction of mental illness associated with unemployment using data
mining. Procedia Computer Science, 177, 556–561. doi:10.1016/j.procs.
2020.10.078.
[87] Kaur I., Doja M., and Ahmad T. (2020). Time-range based sequential mining
for survival prediction in prostate cancer. Journal of Biomedical Informatics,
110, 103550. doi:10.1016/j.jbi.2020.103550.
[88] Zhang Q., Lian B., Cao P., Sang Y., Huang W., and Qi L. (2020). Multi-
source medical data integration and mining for healthcare services. IEEE
Access, 8, 165010–165017. doi: 10.1109/ACCESS.2020.3023332.
 Index

accelerometers 286 support in clinical decisions 106

active and passive attacks types 11 examples of 106–8
AD8232 115, 117, 119 healthcare data sources 95–6
AdaBoosting algorithm 315–16 legal and ethical obstacles of
adaptive slot adaptive frame (ASAF)- artificial intelligence-driven
ALOHA protocol 41 healthcare 96–8
adaptive sparse representation (ASR) major disease areas 103–4
128 types of healthcare data 98–100
ADAS mechanisms equipping modern structured data 98
vehicles 20 unstructured data 98–100
addressing schemes 2 artificial intelligence (AI), for
advanced driver-assistance systems sustainable e-Health 139–52
(ADAS) 20 artificial intelligence and area of
AI algorithms 230 applications 149–51
AI-enabled IoMT 8 control 150
allergic rhinitis 291 diagnosis 150
AlphaZero 97, 99 drug design 150–1
American Hospital Association 55 interpretation 150
ANH-USA 146 monitoring 150
apnea detection 277–9 treatment planning 150
apnea detection algorithm 279 barriers in EHR interoperability 143
application plane 300–1 barriers in EHR system 141–2
artificial intelligence (AI), in broadening health-care facilities at
healthcare 93–108 home during COVID-19
AI techniques for structured and pandemic 147
unstructured data 100–3 e-Health and AI 146
machine learning 100–1 EPRs, EHRs, and clinical systems
natural language processing 141
102–3 futuristic research directions 148
neural networks 101–2 interoperability issues 142–3
applications of 104–6 MNC/MNE characteristics 144–5
development and discovery of reduction in margin of error in
drug 105–6 healthcare 140–1
robotics and artificial intelligence- role of AI in diabetes care 148–9
powered devices 106 sustainability in staff and clinical
solutions based on genetics 105 practice during pandemic 147
356 The Internet of medical things

sustainable development groups smart governance 335

147–8 smart living 336
sustainable healthcare 146 smart mobility 335
sustainable healthcare in aftermath smart people 334
of COVID-19 147 big data method 36
UK-NHS model, characteristics and biomedical engineering 35
enhancements 143–4 biosensors 286
UK-NHS model or MNC blockchain 111, 304
organizational model 145–6 blockchain module, 116
artificial intelligence-driven healthcare blockchain technology 189
legal and ethical obstacles of 96–8 blood coagulation testing 45
artificial intelligence-powered blood pressure monitoring 253, 291
devices 106 blood temperature monitoring 253
artificial neural network (ANN) bluetooth 42, 45, 47, 111, 156, 287,
149–50, 314 294, 312
asthma 45 bluetooth-enabled blood labs 44–5
attacks on IoT 10–11 body mass index (BMI) 71
attribute-based encryption (ABE) 113 body scanning 289
auto patient access 290 body sensor network 5, 294
average motion signal 276 body temperature 291
monitoring 253
big data analysis 234 breast cancer (BC) 155
big data analytics and data mining, for predictive method of 164
healthcare and smart city breathing-related movements 277
applications 331–46
advances in healthcare sector 342–5 camera sensors 22–3
big data for healthcare 343–4 carcinoma melanoma detection 291
data mining for healthcare 345 chronic diseases, management of 4
computational infrastructures for chronic kidney disease dataset 244
smart cities big data analytics MLT details 246
336–40 circuit diagram of system 120
big data based on machine clinical data 98
learning 339–40 cloud-based personal health record
cloud computing 338 system 112
edge computing 339 cloud-based temperature monitoring
fog computing 338–9 system 7
mining methods for big data 340–2 cloud computing 1, 299, 338
another mining method 342 code-based cryptography 211
classification 340–1 cognitive IoMT (CIoMT) 8–9
clustering 341–2 communication stack 2
frequent pattern mining 342 computed tomography (CT), 127, 316
theoretical background of smart computer-aided detection (CADe)
cities 333–6 system 317
smart economy 334 computer-aided diagnosis (CADx)
smart environment 335–6 system 317
 Index 357

computing machines 146 difference of frames (DoF) 274

confusion matrix 164, 341 differential privacy (DP) 322
connected cancer treatment 45–6 dimensionality reduction algorithm
connected care 290 315
connected inhalers 45 discrete cosine transform (DCT) 127
connected wearables 46 driver–vehicle system 20
constant patient monitoring 290 drug development 105–6
control plane 300–1 drug discovery 105–6, 231
convoluted neural network (CNN) 314 drug management 4, 289, 292
cough and bronchial infections 291 DS18B20 117, 119
COVID-19 1, 7–9, 251
worldwide impact on 60–2 ECG self-monitoring 291
in Brazil 61 ECG sensor 119
in China 60–1 edge computing 298, 339
in India 61 electrical medical records (EMR) 333
in Singapore 61–2 electrocardiogram (ECG) 22, 27, 252
COVID-19 Chronicles 62 electronic-health (E-Health)
COVID-19 Health Literacy Project 76 management 251
COVID-19 pandemic 55–6, 58, 63, 80, electronic health records (EHRs) 112,
87, 146, 215, 263 139–40, 284, 333
CPS-IoT systems 194 EHR interoperability, barriers in
customized remote treatment 290 143
EHR system, barriers in 141–2
data aggregation 258 electronic medical records (EMRs)
data anonymization techniques 321 229
data-based attacks 10 electronic patient records (EPRs) 140
data collection and management 296 Elon Musk 107
Data-Driven Investor 284 entrepreneurship 78
data fusion–enabled approach 5 Equivital EQ02 LifeMonitor sensor
data mining (DM) 331 24, 26
data plane 300 Eulerian video magnification (EVM)
data security and privacy 258 277
datasets details 238 Express.js 122–4
decentralization 189 eye disorders 291
decision tree (DT) 239, 314
deep learning (DL) 93–4, 148, 150, feedforward neural network (FFNN)
232, 331 314
sources of data for 102 Fitbit Flex 255
dermatology 231–2 fitness and activity trackers 288
designing treatment plans and FLIR One Pro LT thermal camera 24,
diagnosis 232 26
diabetes 252 flow sensors 286
diabetes dataset 244 fog computing 5, 298, 338–9
MLT details 244 food desert 70
diabetes mellitus 148 fractal analytics 107
358 The Internet of medical things

frequency-domain analysis 25 health-care solutions using

fusion scheme 131–3 smartphones 255
health cloud 307–8
Gaussian filter 6 Health Empowerment by Analytics,
Gibberish 80 Learning and Semantics
global positioning system (GPS) smart (HEALS) 107
sole 289 Health Fog (HF) 305, 308
glucose control 44 health insurance 290
algorithms 151 Health Insurance Portability and
glucose level sensing 252 Accountability Act (HIPAA)
glucose monitoring 44 265
Google’s DeepMind 107 health sector 111
gradient boosting (GBT) 239 applications of IoT in 5
gradient boosting algorithm 315–16 heart disease dataset 242
Guangdong Province 79 MLT details 242
heart rate (HR) 20
health 4.0 application 310 heart rate monitoring 291
applications and enabling heart rate sensor 118–19
technologies mapping 312–13 heart rate variability (HRV) 19
for health-care professionals heterogeneous driver monitoring 21–3
311–12 camera sensors 22–3
for patients 310–11 wearable and inertial sensors 22
for resource management 312 heterogeneous networks 295
HealthBank 189 high frequency (HF) band, 26
health big data analytics 308 H-IoT sensor networks
Healthcare 4.0 305–22 communication technologies for
application 310–13 293
main technologies of 307–9 edge, fog and cloud for 299
ML-based solutions for 321–2 software-defined networking (SDN)
differential privacy 322 for 300
federated learning 322 Hope Act 76
privacy preservation 321–2 hospital-level analysis 206
objective 309–10 HRV analysis 24–6
healthcare data, types of 98–100 Hyperledger Fabric 114
structured data 98
unstructured data 98–100 IBM’s Watson Oncology 106
health-care dataset 228 image processing 234
healthcare data sources 95–6 image sensors 286
health-care domain imminent health-care solutions 254–5
applications of AI in 231 immutability 189
healthcare industry 95 implantable medical sensors 286–7
health-care informatics, evolution of Indian startup, Haptik 107
AI in 226 inertial measurement units (IMUs) 22
health-care IoT (H-IoT) use cases 289 information and communication
healthcare service 82 technology (ICT), 141
 Index 359

information processing 159 HRV analysis 24–6

information technology 181 in-vehicle monitoring systems 19
Infrared Data Association (IrDA) 294 IoMT-enabled healthcare applications
intelligent techniques application, in 12
health-care sector 225–35 IOTA 121
electronic medical records 229 storing sensed data in 122–3
achievement of AI in healthcare IoT and mobile technology, in virus
231–2 outbreak tracking and
AI in healthcare 230–1 monitoring 251–65
challenges of using AI and benefits 256–7
possible solutions 232–3 data assortment and analysis 257
current status of AI in healthcare simultaneous reporting and
234–5 monitoring 256
future scope 233–4 tracking and alerts 257
reluctant to adopt EMR/ challenges 258
digitalization of medical cost 258
procedures 229–30 data overload and accuracy 258
evolution of AI in health-care data security and privacy 258
informatics 226–7 data privacy and security 264–5
health-care dataset 228–9 digital divide among patients 265
healthcare in India 227–8 lack of information control 264–5
Intel’s venture capital arm 107 evolution of healthcare (pandemic)-
internalization theory 144 based IoT 262–3
Internet of Drones (IoDs) 213 healthcare IoT for virus pandemic
Internet of Health Things (IoHT) 307 management 261
internet of medical things (IoMT) increasing availability of social
1–12 networks 263–4
applications of 3–9 enhancing physician efficiency
role of IoMT during COVID-19 264
7–9 improving efficiency of health
challenges and future directions services 263
11–12 improving patients health
elements connected in 5 condition 263–4
security aspects in 9–11 IoT health-care applications 252–5
security requirements and blood pressure monitoring 253
methods 10 blood temperature monitoring
Internet of Services (IoS) 308 253
Internet of Things (IoT) 35, 111, 251, electrocardiogram monitoring
308 252
general review of 48–9 glucose level sensing 252
systematic review of 50 health-care solutions using
internet of vehicle (IoV) 335 smartphones 255
in-vehicle IoT-oriented monitoring imminent health-care solutions
architecture 23–6 254–5
experimental setup 23–4 medication management 254
360 The Internet of medical things

oxygen saturation monitoring summary of associated research

253 46–50
rehabilitation system 253–4 wireless body area networks 39–41
wheelchair management 254 IoT-based breast cancer monitoring
IoT in healthcare 252 system 155–78
use of IoT in virus outbreak and proposed solution 159
monitoring 259–60 breast cancer prediction proposed
using IoT to dissect an outbreak predictive framework 163–5
259 concept of iTBra 159
using IoT to manage patient care dataset 159
259–60 feature ranking with support
use of mobile apps in healthcare vector machines 160–1
260–1 reforms of SMO algorithm 162–3
mobile report 260 SMO algorithm 161–2
saving human resources 260–1 related works 157–9
IoT architecture 2 study and discussion of experimental
IoT attack types, classification of 11 findings 165
IoT-based biomedical communication preprocessing for dataset 165–7
architecture 39 results of SMO classification
IoT-based biomedical healthcare 167–77
approach 35–51 results of SVM-RFE experiment
examples of IoT-based biomedical 167
healthcare devices 43–50 IoT-based healthcare 291
blood coagulation testing 45 IoT-based healthcare equipment, in
bluetooth-enabled blood labs biomedical approach 36
44–5 IoT-based smart rehabilitation
connected cancer treatment 45–6 systems 253
connected inhalers 45 IoT communication 3
connected wearables 46 IoT-connected contact lenses 46
glucose monitoring 44 IoT healthcare 111–25
IoT-connected contact lenses 46 blockchain to secure 115–24
robotic surgery 46 proposed system 115–18
smartwatch app monitors system implementation 118–24
depression 46 existing models to secure 112–15
future work 51 overview of 111–12
IoT and biomedical healthcare working of proposed blockchain
system interconnection 43 system for 117
IoT-based biomedical IoT health-care network 252
communication architecture IoT Medical Devices 155
38–9 IoT sensor network-based health-care
IoT-based healthcare biomedical market 288
applications 37–8 IoT sensor networks, in healthcare 283
problem statement 43 challenges in implementation of
RFID IoT-based biomedical H-IoT and related research
communication protocol 41–2 294–7
 Index 361

communication technologies for decision tree 314

292–4 dimensionality reduction algorithm
contemporary technologies to 315
overcome challenges on gradient boosting algorithm and
297–301 AdaBoosting algorithm
cloud/fog/edge computing for 315–16
H-IoT sensor networks 297–300 K means 315
software-defined networking for linear regression 313
H-IoT sensor networks 300–1 logistic regression 313
role of 287–92 Naı̈ve Bayes 315
wireless sensor networks (WSN) and random forests 315
Internet of Things (IoT) 285–7 support vector machines 313–14
IT-based smart health-care machine learning for Healthcare 4.0
system 207 316–18
iTBra, concept of 159 vulnerabilities for 318–21
due to data annotation 320
JavaScript 122 in data collection 319
JSON data 124 in deployment phase 320–1
in model training 320
key sensors and applications, in in testing phase 321
healthcare 286 machine learning techniques (MLT)
Kheiron Medical 107 237–49
K means 315 clinical data analysis 238–49
knowledge-based systems (KBS) 150 dataset 1 239–42
K-SVD algorithm 131 dataset 2 242–4
dataset 3 244
lattice-based cryptography 211 dataset 4 244–9
lifestyle assessment 4 related work 238
linear regression (LR) 313 magnetic resonance imaging, 127
Lockout Protocol 84 masked-authenticated messaging
logistic regression (LR) 239, 313 (MAM) 121
long short term memory (LSTM) 5, public mode of 122
314 massive data generation 295
low- and middle-income countries MAX30100 117–19
(LMICs) 55 MedBloc 114
low frequency (LF) band, 26 medical Bot 4
medical cyber-physical systems
machine- and deep-learning-based (MCPS) 307–8
COVID-19 treatment system medical data, types of 228
205 medical data analysis 199, 297
machine-based decision-making 177 medical monitoring mechanism 40
machine learning (ML) 36, 100–1, medical nursing system 4
146, 150, 157, 182, 234 medical tourism 258
machine learning algorithms 313–16 medical waste management 290
artificial neural network 314 medication management 254, 284
362 The Internet of medical things

microelectromechanical systems neonatal disorder detection, video

(MEMS) 286 processing systems for 271–4
Microsoft’s Hanover project 107 multiple sensors 273–4
middleware 1–2, 10 single sensor 273
MIFMSF fusion scheme 131 neural networks (NNs) 94, 101–2
MIFMSF method 136 NodeJS 122
MNC organizational model 151 NodeMCU 120
mobile communication networks (5G) noncommunicable disease (NCD) 63
308 non-invasive psycho-physiological
mobile sensors 21 driver monitoring, through IoT-
model training (MT) 320 oriented systems 19–29
modified spatial frequency (MSF) experimental performance
129–30 evaluation 26–9
MongoDB 122 experimental results 27–9
multi-camera systems 273 HR and HRV data 27
multimodal medical image fusion, operating protocol for data
using sparse representation and collection 26–7
modified spatial frequency future works 29
127–36 heterogeneous driver monitoring
dictionary, learning 129–31 21–3
experimental setup and camera sensors 22–3
analysis 131 wearable and inertial sensors 22
overview of fusion scheme 131–3 in-vehicle IoT-oriented monitoring
simulation results and discussion architecture 23–6
133 experimental setup 23–4
objective analysis of standard HRV analysis 24–6
medical image pairs 135 novel coronavirus (nCoV) 55
results on standard multimodal
medical data sets 133–5 open-source IoT application, for
multinational corporation (MNC) healthcare system 186
model 139 Organization for Economic and Co-
multiple sensory data 118 operation Development
multi-scale transforms (MSTs) 128 (OECD) 77
multivariate post-quantum orthogonal matching pursuit algorithm
cryptography mechanism 212 (OMP) 128
oxygen saturation monitoring
National Centre for Infectious Diseases 253, 291
(NCID) 62
National Health Service (NHS) 107 pandemic “COVID-19” and
National University of Singapore assessment of world health
(NUS) Yong Loo Lin School of management and economics
Medicine 62 55–87
natural language processing (NLP) 94, academy 75–6
100, 102–3, 234, 317 high school graduation 75–6
Naı̈ve Bayes 315 psycholinguistics and literacy 76
 Index 363

healthcare, social and economic active part of community 73

challenges in Bangladesh 81–6 determination 72–3
analysis of basic healthcare 82–3 make surrounding foods 73–5
discussion 85–6 stable, predictable employment
groups with special needs 84–5 76–7
methodology 82 patient care robot 290
precarious living 83–4 patient-centric health-care system, IoT
social distancing 84 network for 184–8
health in clinical system 69–70 patient-centric smart health-care
access to medical- and primary- systems 181–216
care doctor 69–70 artificial intelligence and machine
clinical knowledge 69 learning approaches in IoT for
impact of COVID-19 pandemic on smart health-care system
firms 77–81 196–206
computations and statistical artificial intelligence in smart
review 79 health-care system 199–205
data gathering 79 healthcare, COVID-19, and future
expectations of firms 81 pandemic-related datasets for
statistical results 79–80 smart health-care systems 199
time-saving method 79 machine learning aspects in smart
impact of health on emerging health-care system 205–6
COVID-19 pandemic 66–9 blockchain technology and IoT for
social determinants of health 188–90
67–9 need for blockchain and IoT
impact on societies 57–66 integration 189
constructive response reduction role of Internet of Things in
strategies 64–6 health-care system 189–90
emergency caring 63–4 cybersecurity and IoT for 190–5
impact of social manner on IoT and cybersecurity 190–2
COVID-19 58–9 IoT cyberspace and data handling
impact on healthcare facilities processes for information and
59–60 network security 193–4
impact on international healthcare recent developments in ensuring
facilities based on medical confidentiality, integrity, and
services 63 availability (CIA) properties in
impact on low- and moderate- IoT networks for smart health-
income countries 62 care system 192–3
worldwide impact 60–2 research challenges and future
role of health disease 70–2 direction of IoT in 194–5
food deserts on cardiovascular drone and robotics operation
disease (CVD) 70 management using IoT network
food deserts on hypertension and for smart health-care system
chronic kidney disease 71 213–16
SDOH on obesity 71–2 drone and IoT network for health-
social setting 72–5 care operations 213–14
364 The Internet of medical things

medical robots and recent multivariate cryptography for

developments 213 health-care data and system
recent developments and future 212
directions 215–16 post-quantum cryptography
robot, drone, and IoT network application and research
integrations for health-care challenges for smart health-care
applications 214–15 system 212
Internet of Things (IoT) network supersingular cryptography for
for 184–8 health-care data and system
Industrial IoT (IIoT) for smart 212
health-care system 187 virtualization and IoT in IoT for
IoT and applications for health- smart health-care system
care sector 185–6 206–8
research challenges and future operating system and storage
directions 187–8 virtualization for desktop and
resourceful and resource mobile-based smart health-care
constraint device-based smart applications 206–8
health-care developments 187 patient-level analysis 206
IoT and quantum computing for perceptron 102
smart health-care system personal health records (PHR) 333
208–10 personal protective equipment (PPE) 56
challenges of quantum computing Pima dataset 239
to health-care data and data MLT details 240
security 210 PM 2.5 75
quantum computing and smart polysomnography 272
health-care system 209–10 positron emission tomography (PET),
parallel and distributed computing 127
architecture using IoT network post-quantum cryptography 211
for 195–6 preprocessing dataset 159
cloud computing architectures pressure sensors 286
using IoT network for health- principal component analysis (PCA)
care system 196 128, 313
post-quantum cryptography proposed dictionary learning
solutions for futuristic security approach 129
in smart health-care system proposed health monitoring and
210–12 management system 124
code-based cryptography for proposed health monitoring system
health-care data and system 115
211 proposed system, working of 116–18
hash-based cryptography for protocol-based attacks 10
health-care data and system pulmonary infections 291
211–12 pulse oximeter 118–19, 270
lattice-based cryptography for PySpark 238
health-care data and system clinical data analysis 239
211 Python 113
 Index 365

quality of life (QoL) 332 single-photon emission-computed

quality of service (QoS) 295 tomography (SPECT), 127
quantum algorithms 209 skin allergies/infection 291
quantum computation 209 sleep apnea method 5
quantum support vector machine 210 smart body scanner 289
quantum technology 209 smart cities, components of 337
smart city 262
radio-frequency identification (RFID) smart economy 334
1–2, 35, 41, 294 smart environment 335–6
random forest (RF) 239 smart governance 335
recurrent neural network smart healthcare 3, 12
(RNN) 314 smart health-care applications 206–7
rehabilitation system 4, 253–4, 292 smart health-care system
reinforcement learning 340 AI-based technological aspects for,
remote intervention 4 198
remote medication, through ingestible IIoT for 187
sensors 289 smart health-care system, with
remote patient monitoring 288 virtualization 207
remote retail centers 290 smart living 336
remote robotic surgery 292 smart mobility 335
respiratory rate (RR) 22 smart people 334
RFID-based healthcare system smartwatch app monitors depression
for IoT analytics, in medicine 42 46
Riverbed Modeler 37 SMO (Normalized PolyKernel)
RMSSD values 27–8 classification, tests of 167–73
robotics 106, 234 SMO (PolyKernel) classification, tests
robotic surgery 46, 289 of 173–5
Roche’s technology 45 SMO (Puk) classification, tests of 175
R–R interval 25, 28 SMO (RBF) classification, tests of
175–7
scalability 189, 296 SMO algorithm 161–2
secure data aggregation methods 2 reforms of 162–3
security 189 social determinants of health (SDOH)
security and privacy 296 56, 67–9
seizure 269 domains of 67
detection 274–6 social distancing 66, 69, 82, 84–5
semi-supervised learning 340 social media 59, 332, 335
sensor-based devices 285 software systems 183, 196
sensor-based IoT networks, benefits sparse coefficient estimation 132
of 285 SQUIDs (semiconductor quantum
sensor electronics module (SEM) 23 interface devices) 286
sensor layer, block diagram of 117 SR-based fusion schemes 135
sensor module, 115–16 SR-based methods 133
severe acute respiratory syndrome stationary sensors 21
coronavirus (SARS-CoV-2) 55 structured data 98
366 The Internet of medical things

supersingular cryptography 212 apnea detection 277–9

supervised learning 99–100, 332, 340 performance in 279
support vector classifier (SVC) 239 seizure detection 274–6
support vector machine (SVM) 94, performance in 275–6
160, 313–14 video processing systems for
support vector machine recursive neonatal disorder detection
feature elimination (SVM- 271–4
RFE) 155, 160 multiple sensors 273–4
systemic socioeconomic status single sensor 273
(SES) 55 vital signs monitoring 270–1
virtual care 290
tachogram 25 virtualization technology 206
telehealth 4
tele-health technology 147 wearable and inertial sensors 22
temperature sensor 119 wearable technology 156
temperature sensors 286 web application, retrieving data using
Tencent 107 123
time-domain parameters 25 web application module, 116
transactions in tangle 121 wheelchair management 254
transparency 189 wheelchair navigation 292
typical IoT-based biomedical Wi-Fi 40, 195, 294
healthcare system 38 wireless body area networks
(WBANs) 35, 39–41, 290
UK National Health Services (UK- wireless metropolitan area network
NHS) government model 139 (WiMAX) 195
UK-NHS model, characteristics and wireless sensor networks (WSN) 1,
enhancements 143–4 284
ultra-wideband (UWB) 294 hardware for 2
unstructured data 98–100 Wisconsin Diagnostic Breast Cancer
unsupervised learning 99–100, 232, (WDBC) 155
340 dataset 177
updated infrastructure 295 World Health Organization (WHO) 56
upgraded network technologies 295 wound analysis 291

vaccination 58 Zig-Bee 195

very low frequency (VLF) band 25 Zigbee 287, 293–5
video-based solutions, for newborn
monitoring 269–80