FINE TUNE YOUR ALS MANAGEMENT: CONTINUOUS

MONITORING WITH EMG

A PROJECT REPORT

Submitted by

ARUL PRAVIN R (921721111007)

MEENAKSHI M (921721111023)
SURIYA NARAYANAN K R (921721102162)
SIVARAMAN S (921721102150)
NIRMALA K (921721104092)
RAGAVI S (921721104106)

in partial fulfillment for the award of the degree

of

BACHELOR OF ENGINEERING
and
BACHELOR OF TECHNOLOGY

21UGE710-MULTIDISCIPLINARY PROJECT- PHASE I

SETHU INSTITUTE OF TECHNOLOGY

An Autonomous Institute Affiliated to Anna University, Chennai
PULLOOR, KARIAPATTI – 626115
OCTOBER 2024
 SETHU INSTITUTE OF TECHNOLOGY
An Autonomous Institution

BONAFIDE CERTIFICATE

Certified that this project report “FINE TUNE YOUR ALS MANAGEMENT:
CONTINUOUS MONITORING WITH EMG” is the bonafide work of
“ARUL PRAVIN R (921721111007), MEENAKSHI M (921721111023),
SURIYA NARAYANAN K R (921721102162), SIVARAMAN S
(921721102150), NIRMALA K (921721104092) and RAGAVI S
(921721104106) who carried out the project work under my supervision.

SIGNATURE SIGNATURE
Dr. B. Meenakshisundaram, M.E., Ph.D. Mrs. M. Mathina Kani, M.E., (Ph.D).
COURSE COORDINATOR SUPERVISOR
Professor Assistant Professor
Department of Electrical Engineering, Department of Computer Science
Engineering,
Sethu Institute of Technology, Sethu Institute of Technology,
Pulloor, Kariapatti – 626 115 Pulloor, Kariapatti – 626 115
Virudhunagar district. Virudhunagar district.

Submitted for the Viva – Voce Examination is held on

INTERNAL EXAMINER EXTERNAL EXAMINER

ii
 ABSTRACT

Amyotrophic Lateral Sclerosis (ALS) is a progressive neurological disorder

affecting motor neurons, leading to muscle weakness and degeneration.

Monitoring muscle activity consistently and early on is crucial for understanding

disease progression and treatment effectiveness. A proposed system uses surface

electromyogram (sEMG) sensors to detect electrical signals generated during

muscle contractions. These signals are then analyzed by an ESP32 DEVKIT V1

microcontroller, which has strong computational capabilities and built-in Wi-Fi

and Bluetooth features. In real-time, the microcontroller processes EMG data and

wirelessly transmits the analyzed information to a designated monitoring device

(such as a smartphone or computer). Healthcare providers and caregivers receive

timely updates on the patient’s muscle activity, enabling informed decisions about

treatment strategies and personalized care. By combining EMG sensors,

accelerometer and advanced microcontroller technology, this system enhances

ALS monitoring, allowing proactive intervention and improved quality of life for

individuals with ALS. Machine learning algorithm integrated into this system

enhance its capabilities. By analyzing historical sEMG data, models can predict

disease progression and anticipate symptom onset. Furthermore, by analyzing the

relationship between sEMG data and treatment outcomes, personalized treatment

plans can be developed.

Keywords: Amyotrophic Lateral Sclerosis, ESP32 DEVKIT V1, Electromyogram,

Accelerometer, rehabilitation, Spreadsheets.

iii
 ACKNOWLEDGEMENT

First, We would like to thank GOD the almighty for giving me the talent and
opportunity to complete our project phase-I.

We wish to express our earned great fullness to our honorable Founder and
Chairman, Mr. S. Mohamed Jaleel B.Sc., B.L., for his encouragement extended
us to undertake this project work.

We wish to thank and express our gratitude to our Chief Executive Officer
Mr.S.M. Seeni Mohideen, B.Com., M.B.A., Joint Chief Executive Officer,
Mr.S. M. Seeni Mohamed Aliar Maraikkayar,B.E., M.B.A., and Director
Administration Mrs. S. M. Nilofer Fathima, B.E., M.B.A., and Director-R&D,
Dr. S. M. Nazia Fathima, M.E., Ph.D., for their support in this project.

We would like to thank and express our gratitude to our advisor, Dr.
G.D.Sivakumar, M.E, Ph.D., for providing all necessary for the completion of
the project.

We wish to thank and express our gratitude towards our Principal

Dr. G. D. Sivakumar, M.E., Ph.D., for his support of our project work.

We wish to express our profound gratitude to our course coordinator

Dr.B.Meenakshisundaram, M.E, Ph.D., for granting us the necessary
permission to proceed with our project.

We are immensely grateful to our guide, overseer Dr. M. Mathina

Kani, M.E., (Ph.D.,) for encouraging me a lot throughout the course of the
project. We render our sincere thanks for her support in completing this project
successfully.

We thank our parents, faculty members, supporting staff and friends for their
help extended during these times.

iv
 TABLE OF CONTENTS

CHAPTER NO TITLE PAGE NO

ABSTRACT iii
ACKNOWLEDGEMENT iv
LIST OF FIGURES vii
LIST OF TABLES viii
LIST OF ABBREVIATIONS ix

1 INTRODUCTION 1
1.1 GENERAL 2
1.2 MOTIVATION 2
1.3 OBJECTIVE 3
1.4 SOCIAL RELEVANCE OF THE 4
PROJECT

2 LITERATURE SURVEY 5

3 SYSTEM DESIGN 14
3.1 PROBLEM IDENTIFICATION 14
3.2 PROPOSED SYSTEM 15
3.2.1. COMPARISON WITH 16
EXISTING
METHOD
3.3 DESIGN OF THE PROPOSED 18
SYSTEM

v
 3.3.1 DESIGN CONSIDERATION 19
FOR
PUBLIC HEALTH,
SAFETY AND
ENVIRONMENT

3.4 FUNCTIONAL BLOCK DIAGRAM 20

3.5 WORKING PRINCIPLE OF PROPOSED
SYSTEM 21
3.6 METHODS AND MODERN (IT) TOOLS USED 23

4 REFERENCE 27

vi
 LIST OF FIGURES

FIG.NO TITLE PAGE NO

3.1 Hardware Implementation 15

3.2 Flow diagram of proposed system 16
3.3 Existing Solution (MRI) 17
3.4 User Interface (Home) 19
3.5 Proposed system block diagram 20
3.6 AD9232 Module 21
3.7 Flow diagram of proposed system 23
3.8 User Login Page 25

vii
 LIST OF TABLES

TABLE.NO TITLE PAGE.NO

3.1 Parameter Analysis (Accelerometer) 22

3.2 ESP32 DEVKIT V1 Absolute 24
maximum Ratings

viii
 LIST OF ABBREVIATIONS

ALS - Amyotrophic lateral sclerosis

EMG - Electromyograph

WI-FI - Wireless Fidelity.

SVM - Support Vector Machine

RNN - Recurrent Neural Network

ADC - Analog to Digital Convertor

MRI - Magnetic Resonance Imaging

CSF - Cerebrospinal Fluid.

PD - Parkinson's disease

ML - Machine Learning

ix
 CHAPTER- 1
INTRODUCTION

Amyotrophic Lateral Sclerosis (ALS) a relentless adversary that

progressively dismantles the body’s intricate machinery of motor neurons. For
those grappling with ALS, each muscle contraction whether a subtle flutter or a
pronounced tremor becomes a poignant reminder of their vulnerability. Yet,
within this struggle lies an intersection of science and innovation a place where
hope unfurls its fragile wings.Our proposed system, an elegant marriage of
surface electromyogram (sEMG) sensors and the ESP32 DEVKIT V1
microcontroller, seeks to redefine the landscape of ALS monitoring. Picture these
sEMG sensors as silent sentinels, adhering unobtrusively to the skin. Their
mission? To capture the electrical symphony the nuanced crescendos and
diminuendos as muscles engage and release. These electrical whispers, known as
sEMG signals, harbor vital information.

They reveal the orchestra of motor neurons firing, the intricate

choreography of muscle fibers. And within this delicate dance lies the narrative of
ALS progression. Armed with formidable processing capabilities, the ESP32
receives the raw sEMG signals. Its algorithms dissect the electrical murmurs,
teasing out patterns and anomalies. It’s as if the microcontroller listens intently,
translating muscle twitches into actionable insights. But there’s more: Wi-Fi and
Bluetooth features allow the ESP32 to converse with the outside world as an
emissary transmitting critical updates. Google spreadsheets meticulously log
every twitch, every muscle contraction. Continuous monitoring no gaps, no
lapses. But it’s not merely reactive; it’s predictive. Event detection algorithms
raise alarms when patterns deviate. Early intervention becomes routine, not
desperation. And glimpsing ahead, we envision biofeedback during clinical trials:
a dialogue between patient and machine, muscles responding to cues, progress
measured in electrical whispers.

1
 1.1 GENERAL
In this project we are about to record the electrical activities of the
muscles with respect to the action. The aim is to distinguish the electrical activity
of the muscle between volunteer or control-less potentials(ALS readings). A
surface EMG electrode is placed on the skin with the given adhesives which
picks up the signals beneath the skin and sends it to the EMG module. The
module comes with an amplifier and an ADC that converts the analog signal into
digital signal. Along with the EMG sensor the accelerometer is used to record the
movements made by the user. The potentials recorded at the time of movement
are considered to be normal and otherwise abnormal. The transmission of the data
from the microcontroller will be carried out through wireless communication i.e.
the built in WIFI to the Google service account spreadsheet. A website created
for the patient and physician end is created that displays and detects the last ten
readings for analysis.

1.2 MOTIVATION
The motivation behind our project to develop an advanced monitoring
system for Amyotrophic Lateral Sclerosis (ALS) stems from the significant
impact this neurodegenerative disease has on patients and their families. ALS is
characterized by the progressive degeneration of motor neurons, leading to a
gradual loss of muscle control, speech, and the ability to perform daily activities.
This decline imposes emotional and psychological burdens on both patients and
caregivers. Traditional monitoring methods have often been reactive, leaving
individuals vulnerable to sudden health declines without timely interventions.
This highlights the urgent need for innovative solutions that enhance monitoring
capabilities and improve patient outcomes.
Our inspiration arises from the challenges faced by ALS patients and
their loved ones. Witnessing a loved one struggle with ALS can be
overwhelming. Many caregivers feel helpless due to a lack of real-time insights
needed for informed decisions about treatment and care. Our system aims to
2
bridge this gap by providing continuous monitoring of muscle activity through
surface electromyogram (sEMG) sensors. These sensors capture subtle nuances
in muscle contractions, empowering caregivers with data that deepens their
understanding of the disease’s progression. By visualizing muscle activity, we
hope to reduce feelings of helplessness and encourage proactive engagement in
care strategies.
Moreover, the traditional medical approach to ALS management relies
heavily on subjective assessments and periodic evaluations. This can result in
reactive responses to worsening symptoms, leading to missed opportunities for
timely intervention. Our motivation to develop a predictive monitoring system
seeks to shift this paradigm toward a more proactive, data-driven approach. By
utilizing the ESP32 microcontroller, our system can analyze sEMG signals in real
time, detecting deviations from established muscle activity patterns. This
capability can trigger alerts for early intervention, potentially slowing disease
progression and mitigating complications.
Enhancing patient autonomy is another driving force behind our
initiative. ALS often strips away an individual's ability to perform daily tasks
independently, leading to frustration and a loss of agency. By providing real-time
monitoring data, we hope to empower patients to engage actively in their care.
We envision integrating biofeedback mechanisms during clinical trials, allowing
patients to receive immediate feedback on their muscle activity. This interaction
fosters a sense of control and involvement, significantly enhancing the
psychological well-being of ALS patients.

1.3 OBJECTIVE
The primary objective of this project is to develop a robust and user-
friendly system for continuous monitoring of muscle activity in individuals with
Amyotrophic Lateral Sclerosis (ALS). By leveraging sEMG sensors, an ESP32
microcontroller, and wireless connectivity, the system aims to provide real-time
data on muscle contractions to healthcare providers and caregivers. This
3
information will be used to track disease progression, enable early intervention,
improve treatment decision-making, enhance patient quality of life, and facilitate
research.

1.4 SOCIAL RELEVANCE OF THE PROJECT

The proposed system for continuous monitoring of muscle activity in
individuals with Amyotrophic Lateral Sclerosis (ALS) has significant social
relevance due to its potential to improve the lives of patients and their families.
ALS is a devastating disease that progressively affects motor neurons, leading to
muscle weakness and eventual paralysis. The ability to track disease progression
and detect early signs of symptom worsening can empower patients and
caregivers to make informed decisions about their care, manage expectations, and
plan for the future.
By providing real-time data on muscle activity, the system can help to
reduce the anxiety and uncertainty associated with ALS. Patients and caregivers
can gain a better understanding of their condition and the rate at which it is
progressing, allowing them to adjust their lifestyle and activities accordingly.
Additionally, the system can facilitate communication between patients,
caregivers, and healthcare providers, ensuring that everyone is on the same page
and that appropriate interventions can be implemented in a timely manner.
Furthermore, the system has the potential to contribute to ongoing research
efforts to develop more effective treatments for ALS. By collecting valuable data
on disease progression and treatment responses, researchers can gain insights into
the underlying mechanisms of the disease and identify promising therapeutic
targets. Ultimately, the social relevance of this project lies in its ability to
improve the quality of life for individuals with ALS and their families, while also
advancing our understanding of this debilitating disease.

4
 CHAPTER 2
LITERATURE SURVEY

Abdelouahad Achmamad et al. (2023) explores the use of

electromyography (EMG) signals to identify neuromuscular disorders. The
authors propose a machine learning (ML) approach to analyze EMG signals,
which are used to assess nerve impulses in motor and sensory nerves, as well as
muscles. By leveraging advanced ML algorithms, the study aims to improve the
accuracy and efficiency of diagnosing neuromuscular disorders, which can lead
to better emotional health outcomes for patients. The research highlights the
potential of EMG signals as a versatile tool in various biomedical applications
and emphasizes the importance of accurate diagnosis for effective treatment [1].

Alejandro Acien and colleagues (2024), explores the potential of

analyzing typing patterns for detecting and monitoring motor impairment
progression in ALS patients. Published in Scientific Reports, the study proposes a
new digital tool that leverages keystroke dynamics, the way individuals interact
with keyboards. By analyzing factors like typing speed, accuracy, and rhythm, the
tool aims to provide objective data on fine motor function in ALS patients. This
information can be crucial for healthcare providers to track disease progression,
assess treatment effectiveness, and ultimately improve patient care. The study
highlights the potential of utilizing readily available technology like smartphones
to gain valuable insights into ALS progression and potentially improve the lives
of patients [2].

Agurto et al.(2023) investigates the progression of motor and speech

impairments in patients with Amyotrophic Lateral Sclerosis (ALS). Using deep
learning techniques, the researchers analyzed a large dataset of ALS patients'
motor and speech data collected over time. They developed models that could
accurately predict the progression of these impairments, providing valuable
insights into the disease's trajectory. The findings suggest that motor and speech
5
impairments are closely linked and that early detection of both can be crucial for
effective management of ALS. The study highlights the potential of deep learning
for improving our understanding of ALS and developing personalized treatment
strategies. The researchers found that the progression of motor and speech
impairments in ALS is highly individualized, emphasizing the need for
personalized treatment approaches. Furthermore, the study's findings could
inform the development of early diagnostic biomarkers for ALS, potentially
leading to earlier interventions and improved outcomes for patients [3].

Aicha Mokdad, Sidi Mohammed El Amine Debbal & Fadia Meziani

(2022) explains the potential of EMG signals to aid in the diagnosis of ALS. The
study proposes two methods of analyzing EMG signals: bispectrum and
continuous wavelet transform (CWT). These methods were used to assess the
pathological severity of biceps brachii muscle in ALS patients. Four time and
frequency features were extracted from the EMG signals, and three popular
machine learning algorithms (SVM, LDA, and KNN) were implemented to
differentiate between neuropathy and healthy conditions. The results showed that
the SVM classifier achieved the best performance with an accuracy of 95.8%,
precision of 92.59%, and specificity of 92%. These findings suggest that EMG
signal analysis and feature selection can be a promising approach for the
diagnosis of ALS [4].

Alex Larson, Joshua Herrera, Kiran George, and Aaron Matthews (2021)
presents a low-cost EOG device designed to aid individuals with ALS in
controlling computer tasks. The study, authored by researchers from California
State University, Fullerton, explores the potential of EOG technology to provide a
user-friendly communication tool for those with severe motor impairments. The
device utilizes EOG electrodes to measure eye movements, which are then
translated into cursor control and keyboard commands. The authors describe the
development process, including hardware and software design, as well as user

6
testing and evaluation. The results demonstrate the feasibility of using EOG for
effective communication in individuals with ALS, highlighting its potential to
improve quality of life and enhance independence. The study concludes by
discussing the limitations of the current device and outlining potential areas for
future research and development [5].

Bashford et al. investigates the use of surface electromyography (sEMG)

in the diagnosis, prognosis, and monitoring of amyotrophic lateral sclerosis
(ALS). The authors conducted a systematic review of 42 studies focusing on
sEMG techniques in ALS and found that various analytical techniques, including
motor unit decomposition, motor unit number estimation, and measurements of
neuronal hyperexcitability or neuromuscular architecture, can be used to assess
ALS. Among these techniques, the motor unit number index (MUNIX) is the
most validated method for monitoring disease progression and has been used in
clinical trials. The study highlights the importance of multi-disciplinary
collaboration among clinicians, bioengineers, mathematicians, and
biostatisticians to fully capitalize on the potential of sEMG in ALS research and
clinical practice. Overall, the study demonstrates that sEMG is a promising non-
invasive tool for the evaluation and management of ALS [6].

Doulah, A. B. M. S. U., Fattah, S. A., Zhu, W.-P., & Ahmad, M. O

(2014) proposes two methods for classifying neuromuscular diseases using EMG
signals. The first method extracts high-energy discrete wavelet transform (DWT)
coefficients and the maximum value from the EMG data. The second method
extracts the dominant motor unit action potential (MUAP) from the EMG data
and uses DWT coefficients to extract features. Both methods employ a K-nearest
neighbor (KNN) classifier for classification. The results demonstrate that both
methods achieve satisfactory performance in terms of specificity, sensitivity, and
overall classification accuracy, suggesting that wavelet domain feature extraction
can be a promising approach for neuromuscular disease classification. The results

7
of this study further validate the potential of EMG signal analysis as a valuable
tool for the diagnosis of ALS, providing objective and quantitative measures of
muscle dysfunction [7].

Dontje et al. (2021) presents a comprehensive plan for evaluating the

implementation of a novel e-health intervention for ALS patients. The study aims
to assess the feasibility, acceptability, and effectiveness of the "ALS Home-
monitoring and Coaching" program, which utilizes a mobile app and healthcare
coach to provide personalized support and care to ALS patients. The paper
outlines the key components of the program, including patient enrollment, data
collection, and evaluation measures. The researchers will collect data on patient
demographics, disease characteristics, and outcomes related to quality of life,
caregiver burden, and healthcare utilization. Additionally, qualitative interviews
will be conducted to gather feedback on the program's usability, acceptability,
and impact on patient experiences. A participatory action research approach will
be employed, involving patients and caregivers in the design, implementation,
and evaluation of the program. This collaborative approach aims to ensure that the
intervention is responsive to the needs and preferences of the target population.
The study's findings will contribute to the evidence base on the effectiveness of
e-health interventions for ALS and inform future efforts to improve patient care
and outcomes [8].

Gashi et al. presents a novel approach for utilizing mobile and wearable
sensor data to gain valuable insights into the progression of multiple sclerosis
(MS). The authors collected data from MS patients using a variety of sensors,
including accelerometers, gyroscopes, magnetometers, and heart rate monitors,
capturing a wide range of physiological and movement parameters. The collected
data was then used to train machine learning models capable of predicting the
progression of MS. These models were able to identify subtle changes in the
patients' movements and physiological signals, which may be indicative of

8
disease progression. The authors demonstrated the effectiveness of their approach
by comparing the model predictions to clinical assessments of MS progression.
The findings of this study highlight the potential of mobile and wearable sensors
for improving the monitoring and management of MS. By continuously tracking
changes in patients' movements and physiology, these technologies can provide
valuable information for healthcare providers and researchers. This could lead to
earlier detection of disease progression, more personalized treatment plans, and
improved outcomes for MS patients [9].

José Castro, Michael Swash & Mamede de Carvalho(2022) explore the

potential of the cutaneous silent period (CSP) as a diagnostic tool for assessing
upper motor neuron (UMN) involvement in ALS. The CSP is a brief period of
muscle silence that occurs in response to a painful stimulus, reflecting the activity
of inhibitory pathways in the central nervous system. The authors hypothesized
that changes in the CSP, such as delayed onset latency or reduced duration, could
indicate UMN dysfunction in ALS patients. To investigate this, they compared
the CSP in ALS patients with healthy controls, focusing on its onset latency,
duration, and the extent of muscle suppression. The results revealed that ALS
patients exhibited significantly delayed CSP onset latencies compared to controls,
suggesting impaired UMN function. Furthermore, the degree of CSP delay was
correlated with the severity of clinical UMN signs, supporting the notion that the
CSP can serve as a valuable biomarker for UMN involvement in ALS. These
findings contribute to a growing body of evidence suggesting that
neurophysiological measures, in addition to clinical assessments, can provide
valuable insights into the pathophysiology of ALS and aid in its diagnosis and
management [10].

Kraskov et al. presents a novel approach for the continuous and

unobtrusive monitoring of tremor in Parkinson's disease (PD) patients. The
authors utilize supervised machine learning techniques and wearable sensors to

9
accurately detect and classify tremor episodes, offering a promising solution for
improving disease management and patient quality of life. The study involved the
development of a machine learning algorithm capable of distinguishing tremor
episodes from other movement patterns based on data collected from wearable
sensors. These sensors were strategically placed on the patient's body to capture
relevant physiological signals, such as acceleration and gyroscope data. The
algorithm was trained on a large dataset of labeled tremor and non-tremor
episodes, allowing it to learn the characteristic patterns associated with tremor in
PD patients [11].

Margarida Antunes, Duarte Folgado, Marília Barandas, André Carreiro,

Carla Quintão, Mamede de Carvalho & Hugo Gamboa (2023) developed a novel
feature set to characterize sEMG based on quantitative measurements related to
the surface representation of Motor Unit Action Potentials. They conducted
experiments to assess the relevance of this feature set, both individually and in
combination with conventional features from temporal, statistical, spectral, and
fractal domains. Using a dataset containing sEMG data from 17 ALS patients and
24 control subjects, they validated their machine learning pipeline. Remarkably,
the proposed feature set achieved an F1 score of approximately 81.9% for onset
classification and 83.6% for subject classification, relying solely on features
extracted from the right first dorsal interosseous muscle [12].

Marios Kefalas and Milan Kocha proposes a new method for

automatically classifying electromyography (EMG) data as normal or abnormal.
The proposed method uses automated machine learning techniques, specifically
Auto-Sklearn, to select the best machine learning algorithm and hyperparameters
for the classification task. Auto-Sklearn is a framework that automates the
process of machine learning pipeline creation, including feature engineering,
algorithm selection, and hyperparameter tuning. The proposed method achieved
an accuracy of 97.8% in classifying EMG data as normal or abnormal,

10
demonstrating its potential for accurate and efficient automated classification of
EMG data [13].

Martin R. Turner discusses the importance of the Gold Coast criteria in

diagnosing amyotrophic lateral sclerosis (ALS). The Gold Coast criteria,
developed in 2019, provide a standardized framework for diagnosing ALS based
on clinical features and evidence of upper and lower motor neuron involvement.
The article highlights the role of electromyography (EMG) in supporting the
diagnosis of ALS by detecting abnormal muscle activity and nerve damage. EMG
can be used to identify the presence of both upper and lower motor neuron
dysfunction, which is a key requirement for the diagnosis of ALS according to
the Gold Coast criteria. Overall, the article emphasizes the importance of a
comprehensive evaluation, including clinical assessment and EMG, in accurately
diagnosing ALS and providing timely treatment [14].

Muhammad Umar Khan et al. proposes a new method for classifying

electromyography (EMG) signals to assess neuromuscular disorders. The
proposed method involves empirical mode decomposition (EMD) to decompose
EMG signals into intrinsic mode functions (IMFs), followed by logistic
regression for classification. The EMD technique effectively captures the non-
stationary characteristics of EMG signals, while logistic regression provides a
probabilistic framework for classification. The proposed method achieved an
accuracy of 96.67% in classifying EMG signals from healthy subjects and
patients with neuromuscular disorders, demonstrating its potential for accurate
and efficient assessment of neuromuscular disorders using EMG signals [15].

K. M. Naimul Hassan et al. proposes a novel deep learning method for

identifying amyotrophic lateral sclerosis (ALS) using raw electromyography
(EMG) signals. The proposed method, named ALSNet, is a dilated one-
dimensional convolutional neural network that directly processes raw EMG
signals without requiring hand-crafted feature extraction. ALSNet achieved an
11
impressive overall accuracy of 97.74% in identifying ALS subjects from raw
EMG signals, outperforming other existing methods. This study highlights the
potential of deep learning techniques for accurate and efficient diagnosis of ALS
using EMG signals [16].

R. Norel, M. Pietrowicz, C. Agurto, S. Rishoni, and G. Cecchi presents a

promising approach to early diagnosis of ALS. By extracting acoustic features
from voice recordings, the researchers were able to differentiate between
individuals with ALS and healthy controls with a high degree of accuracy. The
findings suggest that acoustic analysis could be a valuable tool for identifying
individuals at risk for ALS, leading to earlier intervention and potentially
improving outcomes. However, further research is needed to validate these results
in a larger and more diverse population, and to explore the potential of acoustic
analysis in detecting ALS at different stages of the disease [17].

Stegmann et al.(2021) explores the potential of speech analysis to detect

and track bulbar changes in patients with Amyotrophic Lateral Sclerosis (ALS).
The authors conducted a study involving 18 ALS patients, collecting frequent
speech recordings over time. Using machine learning techniques, they developed
models that could accurately identify changes in speech patterns associated with
bulbar impairment. The findings suggest that speech analysis can be a valuable
tool for early detection and monitoring of bulbar changes in ALS, potentially
leading to improved patient outcomes. Furthermore, the study highlights the
potential of remote speech analysis for the management of ALS, as it can be
conducted without requiring patients to visit a clinic. This can be particularly
beneficial for patients living in remote areas or with limited mobility.
Additionally, the findings suggest that speech analysis could be used to monitor
the progression of bulbar impairment in ALS, allowing for timely adjustments to
treatment plans [18].

Sushkova et al. investigates the use of area under the curve (AUC)
12
analysis in the context of Parkinson's disease. The authors employ non-
parametric statistical methods, such as the Mann-Whitney U Test, Wilcoxon
Signed-Rank Test, Kruskal-Wallis Test, and Friedman Test, to compare AUC
values between different groups of patients and healthy controls. These methods
are chosen due to their robustness to non-normal data distributions. The study
finds that AUC analysis can be a valuable tool for exploratory data analysis in
Parkinson's disease research, providing insights into disease progression and
potential biomarkers. Furthermore, the authors demonstrate that the 3D AUC
analysis can offer a more comprehensive understanding of the temporal dynamics
of disease progression compared to 2D analysis [19].

Susan Peters and Karin Broberg explores the metabolic dysregulation

associated with ALS, a neurodegenerative disease affecting motor neurons. The
authors review current research on energy metabolism, glucose metabolism, lipid
metabolism, amino acid metabolism, and other metabolic pathways in ALS. They
find that ALS patients exhibit mitochondrial dysfunction, altered glucose
metabolism, dysregulated lipid metabolism, imbalances in amino acid
metabolism, and potential alterations in other metabolic pathways. These
metabolic changes contribute to neuronal death and muscle atrophy.
Understanding these metabolic dysregulations can provide insights into the
disease pathogenesis and potential therapeutic targets [20].

13
 CHAPTER 3

SYSTEM DESIGN

This system incorporates sEMG sensors, an ESP32 microcontroller, and

machine learning algorithms to provide comprehensive ALS monitoring. sEMG
data is acquired, preprocessed, and features are extracted. Acquired sEMG data is
processed by the ESP32 microcontroller and then transmitted wirelessly to a
designated monitoring device (e.g., smartphone, computer). This data can be
further transferred to a cloud-based platform like Google Sheets for centralized
storage and analysis. By integrating with a web portal designed for ALS patients,
healthcare providers can access and visualize the data in real-time. This enables
them to monitor disease progression, evaluate treatment effectiveness, and
provide personalized care. A trained machine learning model analyzes the
features to predict disease progression, detect early-stage changes, and optimize
treatment strategies.

3.1 PROBLEM IDENTIFICATION

Amyotrophic lateral sclerosis (ALS) poses significant diagnostic

challenges due to overlapping symptoms with other neurological disorders, such
as Parkinson’s disease and various motor neuron diseases. Approximately 10% to
15% of patients initially diagnosed with ALS are later found to have an
alternative condition, resulting in false positives. Conversely, nearly 40% of
individuals with ALS receive a false negative diagnosis, often being told they
have a different disease before ultimately confirming ALS. This misdiagnosis
stems from the gradual onset of symptoms like muscle weakness, cramping, and
difficulty speaking or swallowing, which can be attributed to other illnesses. The
complexity of these overlapping symptoms necessitates comprehensive
neurological assessments, including electromyography (EMG) and nerve
conduction studies, to improve diagnostic accuracy.

14
 3.2 PROPOSED SYSTEM
A proposed system leveraging surface electromyogram (sEMG) sensors
offers a promising alternative for diagnosing and monitoring conditions like
amyotrophic lateral sclerosis (ALS). This innovative approach detects the
electrical signals produced during muscle contractions, providing valuable
insights into muscle function. The sEMG sensors capture real-time data, which is
then processed by a microcontroller that analyzes the electrical activity in the
muscles.

Fig 3.1 Hardware Implementation

Once the data is processed, it is wirelessly transmitted to a designated
monitoring device, such as a smartphone or computer. This allows for continuous
monitoring of the patient’s muscle activity, offering healthcare providers and
caregivers timely updates on any changes or abnormalities. Such real-time
insights can significantly enhance decision-making, enabling healthcare
professionals to tailor treatment strategies based on the patient's current
condition. Additionally, the accuracy of this sEMG-based system is comparable
15
to traditional diagnostic methods like MRI and cerebrospinal fluid analysis,
making it a viable option for monitoring ALS and other neuromuscular disorders.
By providing immediate access to crucial data, this system fosters personalized
care, allowing for timely interventions and adjustments to treatment plans.
Ultimately, this technology aims to improve patient outcomes by enhancing the
accuracy and efficiency of monitoring neuromuscular health. A trained machine
learning model analyzes the features to predict disease progression, detect early-
stage changes, and optimize treatment strategies. Processed data and predictions
are transmitted wirelessly to a monitoring device, and historical data is stored for
analysis.

Fig 3.2 Flow diagram of proposed system

3.2.1 COMPARISON WITH EXISTING METHOD

The existing diagnostic approach for ALS heavily relies on invasive

techniques such as Magnetic Resonance Imaging (MRI) and cerebrospinal fluid
(CSF) analysis. While these methods can provide valuable insights, they come
with certain limitations. MRI, while effective in detecting structural changes in
16
the brain, may not be sensitive enough to detect early-stage ALS or differentiate
it from other neurological disorders. Additionally, CSF analysis, although
informative, can be invasive and may not always yield definitive results. The
process of obtaining CSF can also be uncomfortable for patients and carries
potential risks.

In contrast, our proposed solution utilizes electromyography (EMG) to

assess muscle function and nerve activity. EMG involves placing electrodes on
the skin to record electrical signals generated by muscles. By analyzing these
signals, it is possible to identify abnormalities associated with ALS, such as
muscle weakness, fasciculations (involuntary muscle twitches), and denervation.
EMG is a non-invasive procedure that can be repeated frequently, allowing for
monitoring of disease progression and evaluation of treatment response. This
makes it a more practical and patient-friendly approach compared to MRI and
CSF analysis.

Fig 3.3 Existing Solution(MRI)

Furthermore, our proposed solution integrates machine learning

algorithms to analyze EMG data. These algorithms can identify patterns and
features in the EMG signals that are characteristic of ALS, even in early stages of
the disease. This can help improve diagnostic accuracy and reduce the likelihood

17
of misdiagnosis. By combining EMG with machine learning, we aim to develop a
more sensitive and specific diagnostic tool for ALS.

Limitations of existing system:

The existing diagnostic approach for ALS, relying on MRI and CSF
analysis, faces several limitations. MRI can be less sensitive in detecting early-
stage ALS and may not differentiate it from other disorders. CSF analysis is
invasive and may not always yield definitive results. These limitations can
contribute to misdiagnosis and delays in appropriate treatment. The existing
diagnostic approach for ALS heavily relies on invasive techniques such as
Magnetic Resonance Imaging (MRI) and cerebrospinal fluid (CSF) analysis.
These methods can be time-consuming, expensive, and uncomfortable for
patients. Additionally, they may not always provide definitive diagnoses,
especially in early stages of the disease, leading to delays in treatment and
potential misdiagnosis.

3.3 DESIGN OF PROPOSED SYSTEM:

Our proposed solution aims to address the limitations of the existing
diagnostic approach for ALS by utilizing electromyography (EMG) and machine
learning techniques. EMG is a non-invasive procedure that records the electrical
activity of muscles, providing valuable insights into muscle function and nerve
health. By analyzing EMG signals, we can identify abnormalities associated with
ALS, such as muscle weakness, fasciculations, and denervation.To enhance the
accuracy and sensitivity of ALS diagnosis, we propose integrating machine
learning algorithms into the EMG analysis process. These algorithms can identify
patterns and features in the EMG signals that are characteristic of ALS, even in
early stages of the disease. By training machine learning models on large datasets
of EMG recordings from ALS patients and healthy controls that can assist in
differentiating ALS from other neurological disorders.

18
 Fig 3.4 User Interface (Home)
Surface electrodes will be placed on specific muscle groups to record
EMG signals. The data will be collected at a high sampling rate to capture subtle
changes in muscle activity. The acquired EMG data will undergo preprocessing to
remove noise, artifacts, and baseline drift. This will ensure that the data is clean
and suitable for analysis. Relevant features will be extracted from the
preprocessed EMG signals. These features may include time-domain features
(e.g., mean, standard deviation, root mean square) and frequency-domain features
(e.g., power spectrum, spectral entropy). A machine learning model will be
trained on a dataset of labeled EMG recordings. The model will learn to classify
EMG signals as either from ALS patients or healthy controls. The trained model
will be used to classify new EMG recordings as either ALS or non-ALS.

3.3.1 DESIGN CONSIDERATION FOR PUBLIC HEALTH, SAFETY

AND ENVIRONMENT

When designing any product, system, or infrastructure, it is essential to

prioritize public health, safety, and environmental protection. These
considerations are paramount to ensure the well-being of individuals and
19
communities, as well as the sustainability of our planet. By designing products
that are accessible to all, safe to use, and promote healthy behaviors, we can
contribute to public health. Additionally, incorporating safety features,
conducting risk assessments, and adhering to safety standards can help prevent
accidents and injuries. By considering these factors, we can create products and
systems that not only serve their intended purpose but also contribute to the well-
being of individuals and communities, and protect the environment for future
generations.

3.4 FUNCTIONAL BLOCK DIAGRAM

The functional block diagram illustrates a system designed to process
data from two sensors: an sEMG (surface Electromyography) sensor and an
accelerometer. Both sensors feed data into a microcontroller, specifically the
ESP32 Devkit V1, which serves as the central processing unit. The
microcontroller processes the incoming data and subsequently transmits it to a
spreadsheet for storage and analysis. The integration of these components enables
efficient data collection and analysis, making it suitable for applications in health
monitoring, rehabilitation, or fitness tracking. Overall, the diagram represents a
streamlined approach to collecting and utilizing sensor data, highlighting the
interconnectedness of hardware and software in modern data-driven applications.

Fig 3.5 Proposed system block diagram

20
 3.5 WORKING PRINCIPLE OF PROPOSED SYSTEM
Here the proposed system consists of an ESP32 DEVKIT V1
microcontroller as the main processor. The controller has WIFI and bluetooth
features and fulfills our project requirements to demonstrate our proposed system.
Our proposed system consists of a power source, microcontroller, surface EMG
sensor, ADXL345 accelerometer module and a button.

In this solution, the surface EMG sensor is recognized as a crucial sensor

to record the electrical activity of muscles. These sensors are typically placed on
the skin above the muscle of interest, where they detect the electrical signals
generated by muscle contractions.

The data from these sensors can be processed by an Arduino

microcontroller to analyze muscle activity, monitor muscle fatigue, and control
prosthetic devices or exoskeletons. Common surface EMG sensors have
specifications such as a sampling rate of 1000 Hz, a sensitivity of 1 mV/V, and a
bandwidth of 20-500 Hz.

Fig 3.6 AD9232 Module

21
 The AD8232 sensor module is a precision instrumentation amplifier
designed to amplify low-level signals from sensors. It features a high gain, low
noise, and wide bandwidth, making it ideal for a variety of applications, including
biopotential measurements, strain gauge conditioning, and sensor signal
amplification. The AD8232 module typically has a gain range of 1 to 1000, a
noise level of 0.5 nV/Hz, and a unity-gain bandwidth of 1 MHz.

EMG sensor electrode adhesives are used which are essential for
ensuring a secure and reliable connection between the electrode and the skin.
These adhesives help to minimize skin irritation and artifact noise, which can
interfere with the accuracy of EMG measurements.

Common types of electrode adhesives include conductive gel, dry

electrodes, and self-adhesive electrodes. The choice of adhesive depends on
factors such as the duration of the recording, the skin sensitivity of the patient,
and the specific requirements of the application.

Table 3.1 Parameter Analysis (Accelerometer)

PARAMETER MIN TYPE MAX UNITS

Clock cycle time 10 - 1000 us
Start hold time 0.6 - 400 ns
Data hold time
0 - 400 ns
(SDAOUT pin)
Data setup time 100 - 400 ns
Start condition time 0.6 - 400 ns
Stop condition time 0.6 - 400 ns
Rise time for data - - 1 us
Fall time for data - - 1 us
Idle time start 1.3 - 400 ns

22
 3.6 METHODS AND MODERN (IT) TOOLS USED
The system involves strategically placing sEMG sensors on the
patient's muscles to capture electrical signals generated during contractions.
These signals are then amplified, digitized, and processed by the ESP32
DEVKIT V1 microcontroller.

Fig 3.7 Flow diagram of proposed system

The microcontroller employs signal processing techniques to remove

noise, enhance signal quality, and extract relevant features that characterize
muscle activity. A machine learning algorithm, such as a recurrent neural
network (RNN) or a support vector machine (SVM), is integrated into the
system to analyze the processed sEMG data. The algorithm is trained on a
dataset of sEMG recordings from ALS patients and healthy individuals to learn
patterns associated with disease progression and muscle function. The
processed data and machine learning predictions are transmitted wirelessly to a
designated monitoring device (e.g., smartphone, computer) via Wi-Fi or
Bluetooth, allowing healthcare providers and caregivers to monitor the patient's
muscle activity in real-time and make informed decisions about their care.
The sEMG sensors and ESP32 microcontroller can be integrated into
wearable devices, such as wristbands or armbands, for more convenient and
23
unobtrusive monitoring. The ESP32's built-in Wi-Fi and Bluetooth capabilities
enable seamless integration with other IoT devices and systems, allowing for
more comprehensive monitoring and data analysis. Advanced data analytics
techniques, such as data mining and visualization, can be employed to extract
meaningful insights from the collected sEMG data. This can help identify
trends in disease progression, evaluate the effectiveness of treatments, and
develop personalized care plans.
A cloud-based platform can be implemented to facilitate data storage,
analysis, and sharing. This platform can provide secure storage for sEMG data,
enable remote access for healthcare providers, and facilitate collaboration
among different stakeholders.

Table 3.2: ESP32 DEVKIT V1 Absolute maximum ratings

Operating Minimum Maximum Units

conditions
Supply voltages
VDD 3.0 3.6 V
VSS 0 V
Input voltage
VI 0 3.3 V
Output voltage
VO VSS to VDD VSS to
VDD
Total Power
Dissipation
PD (TA=85 C) 200 500 mW
Temperatures
Operating -40 +85 C
Temperature
Storage -40 +85 C
Temperature

The system processes the EMG data collected from electromyography

24
(EMG) sensors, which is then transmitted to a software backend for analysis.
Upon receiving the digitized signals from the ESP32 microcontroller, the
software applies signal preprocessing techniques, including noise reduction and
normalization, to ensure clean and reliable data. Feature extraction algorithms are
then employed to identify essential characteristics of the EMG signals, such as
amplitude, frequency content, and signal patterns, which serve as the input for the
machine learning model.

Fig 3.8 User Login Page

A user login page is created for creating a database for each and every
user separately along with their demographics, like age, gender, hereditary
abnormalities and allergic conditions. After the webpage creation and integration
with the spreadsheet, a machine learning algorithm is integrated into the system
to analyze these features. The algorithm is trained on datasets containing EMG
recordings from both ALS patients and healthy individuals, allowing it to identify
patterns related to muscle activity and disease progression. After the analysis, the
results are calculated, and the output, including any insights or predictions
regarding the patient's muscle activity, is rendered on a web-based platform.
25
 The data is made accessible to authorized healthcare providers or
caregivers via a secure, authenticated system, allowing real-time monitoring of
the patient's condition and facilitating informed decision-making regarding
treatment and care. The software can also be extended to cloud-based platforms
for data storage, sharing, and remote access, ensuring seamless integration.

26
 CHAPTER 4
REFERENCES

1. L.I. Grad, G.A. Rouleau, J. Ravits, N.R. Cashman, Clinical spectrum of

amyotrophic lateral sclerosis (ALS), Cold Spring Harb. Perspect. Med. 7 (8)
(2021) a024117, http://dx.doi.org/10.1101/cshperspect.a024117.

2. E. Longinetti, F. Fang, Epidemiology of amyotrophic lateral sclerosis, Curr.

Opin. Neurol. 32 (5) (2019) 771–776, http://dx.doi.org/10.1097/WCO.00730.

3. R.B. Daroff, M.J. Aminoff, Encyclopedia of the Neurological Sciences,

second ed., Elsevier Science, Academic Press, 2019.

4. E. Tiryaki, H.A. Horak, ALS and other motor neuron diseases, Contin.
LifelongLearn.Neurol.20(5)(2020)1185–1207,http://dx.doi.org/10.1212/01.C
ON.0455886.14298.a4.

5. J. Duchateau, R.M. Enoka, Human motor unit recordings: Origins and insight
into the integrated motor system, Brain Res. 1409 (2021) 42–61,
http://dx.doi.org/10.1016/j.brainres.2021.06.011.

6. C.K. Thompson, F. Negro, M.D. Johnson, M.R. Holmes, L.M. McPherson,

R.K. Powers, D. Farina, C.J. Heckman, Robust and accurate decoding of
motoneuron behaviour and prediction of the resulting force output, J. Physiol.
596 (14) (2018) 2643–2659, http://dx.doi.org/10.1113/JP276153.

7. B.R. Brooks, R.G. Miller, M. Swash, T.L. Munsat, El escorial revisited:

Revised criteria for the diagnosis of amyotrophic lateral sclerosis, Amyotroph.
Lateral Scler. Other Mot. Neuron Disord. 1 (5) (2023) 293–299,
http://dx.doi.org/10.1080/146608202300079536.

8. M. de Carvalho, R. Dengler, A. Eisen, J.D. England, R. Kaji, J. Kimura, K.

Mills, H. Mitsumoto, H. Nodera, J. Shefner, M. Swash, Electrodiagnostic
criteria for diagnosis of ALS, Clin. Neurophysiol. 119 (3) (2018) 497–503,
http://dx.doi.org/10.1016/j.clinph.2018.09.143.

9. X. Zhang, P.E. Barkhaus, W.Z. Rymer, P. Zhou, Machine learning for

supporting diagnosis of amyotrophic lateral sclerosis using surface
electromyogram, IEEE Trans. Neural Syst. Rehabil. Eng. 22 (1) (2019) 96–
103, http://dx.doi.org/10.1109/TNSRE.2019.2274658.

10. D. Farina, F. Negro, M. Gazzoni, R.M. Enoka, Detecting the unique

representation of motor-unit action potentials in the surface electromyogram,
J.Neurophysiol.100(2018)1223–1233,http://dx.doi.org/10.1152/jn.90219.2018

27
11. G.D. Meekins, Y. So, D. Quan, American association of neuromuscular &
electrodiagnostic medicine evidenced-based review: Use of surface
electromyography in the diagnosis and study of neuromuscular disorders,
Muscle Nerve 38 (4) (2019) 1219–1224, http://dx.doi.org/10.1002/mus.21055.

12. J. Bashford, K. Mills, C. Shaw, The evolving role of surface

electromyography in amyotrophic lateral sclerosis: A systematic review, Clin.
Neurophysiol.131.(4)(2020),942–950,
http://dx.doi.org/10.1016/j.clinph.2019.12.007.

13. M. Hakonen, H. Piitulainen, A. Visala, Current state of digital signal

processing in myoelectric interfaces and related applications, Biomed. Signal
Process. Control 18 (2019) 334–359.

14. C. Spiewak, A comprehensive study on EMG feature extraction and

classifiers, Open Access J. Biomed. Eng. Biosci. 1 (1) (2018) 17–26,
http://dx.doi.org/10.32474/OAJBEB.2018.01.000104.
15.L.R. Quitadamo, F. Cavrini, L. Sbernini, F. Riillo, L. Bianchi, S. Seri, G.
Saggio, Support vector machines to detect physiological patterns for EEG and
EMG-based human–computer interaction: a review, J. Neural Eng. 14 (1)
(2017) 011001, http://dx.doi.org/10.1088/1741-2552/14/1/011001.

16.S. Anowarul Fattah, Identifying the motor neuron disease in EMG signal
using time and frequency domain features with comparison, Signal Image
Process.An.Int.J.3(2)(2022)99–114, http://dx.doi.org/10.5121/sipij.2022.3207.

17.A. Verma, B. Gupta, Detecting neuromuscular disorders using EMG signals

based on TQWT features, Augment. Hum. Res. 5 (1) (2020) 8,
http://dx.doi.org/10.1007/s41133-019-0020-7.

18.A. Mokdad, S.M. El Amine Debbal, F. Meziani, Diagnosis of amyotrophic

lateral sclerosis (ALS) disorders based on electromyogram (EMG) signal
analysis and feature selection, Polish J. Med. Phys. Eng. 26 (3) (2020) 155–
160, http://dx.doi.org/10.2478/pjmpe-2020-0018.

19.A. Belkhou, A. Achmamad, and A. Jbari, “ Classification and Diagnosis of

Myopathy EMG signals using the Continuous Wavelet Transform,” 2019 Sci.
Meet. Electr. Biomed. Eng. Comput. Sci., pp. 1–4, 2020.

20.R. Norel, M. Pietrowicz, C. Agurto, S. Rishoni, and G. Cecchi, “Detection of

Amyotrophic Lateral Sclerosis (ALS) via Acoustic Analysis,” in Proc.
Interspeech, pp. 377-381, 2018.

28