Healthcare Chatbot

A major project report submitted in partial fulfilment of the

requirement for the award of degree of

Bachelor of Technology
in
Computer Science & Engineering / Information Technology
Submitted by
Amisha Chauhan (201117)
Naman Puri (201318)

Under the guidance & supervision of

Dr. Aman Sharma
Asst. Prof. (SG)

Department of Computer Science & Engineering and

Information Technology
Jaypee University of Information Technology, Waknaghat,
Solan - 173234 (India)
 CERTIFICATE

This is to certify that the work which is being presented in the project report titled “Healthcare
Chatbot using NLP Techniques” in partial fulfilment of the requirements for the award of the
degree of B.Tech in Computer Science and Engineering and submitted to the Department of
Computer Science and Engineering, Jaypee University of Information Technology, Waknaghat
is an authentic record of work carried out by “Amisha Chauhan 201117” and “Naman Puri
201318” during the period from August 2023 to May 2024 under the supervision of Dr. Aman
Sharma, Department of Computer Science and Engineering, Jaypee University of Information
Technology, Waknaghat.

Amisha Chauhan Naman Puri

Roll No.: 201117 Roll No.: 201318

The above statement made is correct to the best of my knowledge.

i
 Candidate’s Declaration
We hereby declare that the work presented in this report entitled ‘Healthcare Chatbot’ in partial
fulfilment of the requirements for the award of the degree of Bachelor of Technology in Computer
Science & Engineering / Information Technology submitted in the Department of Computer
Science & Engineering and Information Technology, Jaypee University of Information Technology,
Waknaghat is an authentic record of my own work carried out over a period from August 2023 to
May 2024 under the supervision of Dr. Aman Sharma (Asst. Prof. (SG), Department of Computer
Science & Engineering and Information Technology).

The matter embodied in the report has not been submitted for the award of any other degree or
diploma.

(Student Signature with Date) (Student Signature with Date)

Student Name: Amisha Student Name: Naman Puri
Chauhan Roll No.: 201117 Roll No.: 201318

This is to certify that the above statement made by the candidate is true to the best of my
knowledge.

(Supervisor Signature with Date)

Supervisor Name: Dr. Aman Sharma
Designation: Asst. Prof. (SG)
Department: Computer Science & Engineering and Information
Technology Dated:

ii
 ACKNOWLEDGEMENT

Firstly, we express our heartiest thanks and gratefulness to almighty God for His divine blessing
making it possible to complete the project work successfully.
We are grateful and wish my profound indebtedness to Supervisor Dr. Aman Sharma,
Assistant Professor, Department of CSE Jaypee University of Information Technology,
Waknaghat. Deep Knowledge & keen interest of my supervisor in the field of “Cloud
Computing” to carry out this project. His proper guidance, encouragement, constant
supervision, constructive criticism, valuable advice, reviewing our work and correcting them at
all stages have made it possible for partial completion of this project.
We would also generously thank each one of those individuals who have helped us
straightforwardly or in a roundabout way in making this project a win. In this unique situation,
We might want to thank the various staff individuals, both educating and not instructing, which
have developed their convenient help and facilitated our undertaking.

Amisha Chauhan
201117

Computer Science & Engineering and Information Technology

Jaypee University of Information Technology, Waknaghat

Naman Puri
201318

Computer Science & Engineering and Information Technology

Jaypee University of Information Technology, Waknaghat

iii
 TABLE OF CONTENT

CONTENT PAGE NO.

Chapter 1: Introduction

1.1) Introduction 1

1.2) Problem statement 2

1.3) Objective 2

1.4) Significance and motivation of the project 2

1.5) Organization of Project Report 4

Chapter 2: Literature Review

2.1) Overview of Relevant Literature 5

2.2) Key Gaps in the Literature 18

Chapter 3: System Development

3.1) Requirements and Analysis 19

3.2) Project Design and Architecture 23

3.3) Data Preparation 25

3.4) Implementation 30

3.5) Key Challenges 38

iv
Chapter 4: Testing

4.1) Testing Strategy 39

4.2) Test Cases and Outcomes 40

Chapter 5: Results and Evaluation

5.1) Results 43

5.2) Comparison with Existing Solutions 45

Chapter 6: Conclusions and Future Scope

6.1) Conclusion 47

6.2) Future Scope 48

References 50

v
 LIST OF FIGURES

Figure Caption Page

No. No.
Fig 2.1.1 Literature surveys consisting of tools, results and limitations observed 17

Fig 3.2.1 This is a rundown of the process of our project, the steps are 24
implemented to ensure a smooth implementation.

Fig 3.3.1 Description of total instances and symptoms available in the dataset 25

Fig 3.3.2 Description of unique number of diseases, symptoms and precautions, 25

severity of symptoms is also a key to alarm the patient

Fig 3.3.3 Description of instances available in various phases of the dataset. 26

Fig 3.3.4 The above figure of our dataset shows various diseases and their 28
corresponding symptoms. This helps our model to match the disease
with likely symptoms.

Fig 3.3.5 The diseases are then binary encoded which makes it easier to train the 29
model as working on binary digits is way easier.

Fig 3.4.1 List of all the pre included libraries and packages in python which we 30
have made use of.

Fig 3.4.2 The dataset is divided into two categories: training data and testing data 31
to help in better performance.

Fig 3.4.3 Configuration file of our model which lists down various frameworks 32
and their versions to run our code.

Fig 3.4.4 The data is then trained where the disease is matched with the 32
symptoms.

vi
Fig 3.4.5 The user is able to list their symptoms and the chatbot responds back 33
with the likely disease.

Fig 3.4.6 This is the interface which users will be looking at when they first boot 33
the chatbot.

Fig 3.4.7 This figure shows the interaction between the chatbot and the user. 34

Fig 3.4.8 Symptom precaution. 36

Fig 3.4.9 Symptoms severity ranging from 1-7. 37

Fig 3.4.10 Symptoms description dataset. 38

Fig 4.2.1 The user is interacting with the chatbot and replying to the queries. 41

Fig 4.2.2 Another user wishes to interact with the chatbot and the corresponding 42
disease is predicted.

Fig. 5.1.1 Result of the working model when data is fed to it. 44

vii
 ABSTRACT

Healthcare is rapidly moving toward patient-centred care, and technology has been essential in
developing the quality of personalised healthcare. This is a report about an artefact, which is a
healthcare chatbot that utilises state-of-the-art natural language processing technology. Chatbot
was developed with an aim of enhancing patient participation, provisioning relevant messages
in time, as well building linkage between doctors and patients.

It leverages highly advanced NLP systems that can understand the sentiments of conversations
and respond back as a real human would, making communication easy and relevant within its
context. It can be integrated in existing healthcare platforms and hence made accessible by
patients for queries such as health information to specific medical inquiries. Through artificial
intelligence or machine learning algorithms, it is possible for the chatbot to keep improving on
its performance adjusting according to the ever-changing language and user’s preferences.

The healthcare chatbot's key features include symptom analysis, prescription reminders, and
health-related FAQs. By adhering to healthcare standards and using encryption mechanisms for
sensitive information, the system assures data security and privacy compliance.

This study adds to the expanding body of work in harnessing natural language processing (NLP) for
healthcare applications by providing insights into the creation, deployment, and assessment of a
chatbot targeted at enhancing patient experiences and optimising healthcare communication
channels. The findings highlight the potential for NLP-based chatbots to play an important role in
the future of healthcare by enhancing accessibility, efficiency, and patient-centred treatment.

viii
 CHAPTER 1: INTRODUCTION

1.1 Introduction
To fulfil the rising need for accessible and efficient patient services in the quickly expanding
healthcare scene, the integration of cutting-edge technology has become critical. Natural
Language Processing (NLP) has emerged as a transformational force, revolutionising the
way people engage with healthcare providers. This study describes a revolutionary
healthcare chatbot system that uses NLP techniques to improve patient engagement and
communication in the healthcare area.

The healthcare system is by no means perfect. The Indian healthcare system is faced with
several challenges being increased healthcare costs, need for nursing, care for senior
citizens, poor not being able to afford healthcare and negligence of public health due to lack
of utilisation of adequate funds. [1] Only 37% of the total Indian population is covered by
health insurance as of 2022, which indicates that close to a billion are still hanging onto
mere hope to afford when an emergency may arrive.

The chatbot, designed to improve user experience, employs powerful NLP models to grasp
and create contextually relevant replies, mimicking normal human interaction. This
intelligent system is intended to help with a variety of healthcare-related tasks, such as
symptom assessments and the transmission of relevant health information.

With the healthcare sector moving towards a paradigm that is more patient-centric, this
research attempts to add to the increasing amount of information about the use of NLP in
healthcare settings. This study aims to highlight the potential influence of natural language
processing (NLP) technologies on healthcare communication channel optimisation and,
eventually, patient outcome improvement by offering a thorough overview of the healthcare
chatbot's development and deployment.

1
1.2 Problem Statement
An enormous need for easily accessible and effective patient assistance systems exists in the
healthcare industry. Frequently, patients need immediate access to medical information and
basic diagnostic advice. Using cutting-edge tools like Scikit-Learn, Pandas, and Pyfiglet, the
task is to develop a system that elegantly blends these functions.

The need to create a healthcare chatbot that makes use of these technologies to give
individualised healthcare support and provide correct medical information is outlined in
this issue statement. This technology may address significant difficulties in the
healthcare industry and improve the patient experience through accurate information
transmission and personalised replies.

1.3 Objectives
● To study the existing technologies for a chatbot

● To propose framework for a chatbot

● Testing and validation of the proposed framework

● We aim to create a platform which provides textual, verbal and visual facilities to the

user for self-diagnostics of health.

1.4 Significance and motivation of the project

This initiative bears relevance as it successfully bridges the gulf between sophisticated
technology and healthcare, solving major sectoral concerns. The usage of Natural Language
Processing (NLP) in healthcare chatbots is spurred by essential reasons, including greater
communication, simplified processes, and improved accessibility to medical information,
together contributing to the evolution of healthcare services.

2
● Enhanced Patient Engagement: The project's goal is to encourage patients to take a
more educated and proactive part in their treatment. Patients may have meaningful
conversations about their health, symptoms, and treatment plans with a chatbot that can
comprehend and answer natural language inquiries. This fosters a sense of empowerment
and engagement in patients' care.

● Effective Information Distribution: By providing prompt answers to user questions,

the healthcare chatbot acts as a trustworthy information source. This effective
information-dissemination process helps to decrease false information and increase user
health literacy in addition to increasing the accessibility of healthcare knowledge.

● Optimised Healthcare Processes: The chatbot expedites standard healthcare procedures by

providing multiple features. By reducing administrative responsibilities and increasing the
efficiency with which healthcare resources are used, this optimisation frees up healthcare
practitioners to concentrate on more intricate and important activities.

● 24/7 Support and Accessibility: The chatbot offers consumers a continuous support
system and is accessible around-the-clock. This feature is especially important when it
comes to medical situations since it guarantees that people may get advice and
information whenever they need it, which might lessen the load on emergency services.

● Data-Driven Insights: An abundance of useful data on user interactions, preferences,

and healthcare queries is produced by the project. Insights into user behaviour, public
health trends, and areas where healthcare services may be further enhanced can be gained
from analysing this data, which supports evidence-based decision-making in healthcare
management.

● Technology Advancement in Healthcare: The initiative is in line with the larger trend
of technology improvement in healthcare by utilising NLP methods. It demonstrates how
cutting-edge technologies may improve patient care, communication, and the provision
of healthcare services.

3
1.5 Organization of Project Report

Chapter 1: An overview of the project that includes information about its purpose, its goals,
and how it will benefit users is provided.

Chapter 2: Survey of the literature for the project. This covers the several project reports on
the healthcare-focused chatbot programmes and their uses.

Chapter 3: System Development: In this section, we have covered the project's primary
design, the connections between its components, the platforms it was built on, and a summary
of the algorithms it employed.

Chapter 4: Model testing is completed, ensuring functionality, accuracy, and performance

meet specified requirements before deployment in the production environment.

Chapter 5: The model is executed, generating results that are analysed to confirm the expected
output and validate performance in real-world scenarios.

Chapter 6: In conclusion, we have discussed the project's results, its future scope, and what
more may be done right now.

4
 CHAPTER 2: LITERATURE REVIEW

2.1 Overview of Relevant Literature

In this section we have looked at the past research and studies around chatbot and its
implementation. Variety of such researches have been carried out and their target causes have
varied as well. We shall also look at the various tools used by them so a comparison can be
drawn between our implementation to that of theirs.

Anuj Patil [2] created a healthcare chatbot using AI. In the process, he used the following
techniques: NLP, Google API, Rasa NLU and AWS services. The key point in his research was
his use of a large dataset to help with better accuracy.

In another research [3], author Greeshma Kurup created an AI Conversational Chatbot for
Primary Healthcare Diagnosis. She used NLP and concepts of deep learning. She employed
decision tree classifiers in her methodology. Her flaw was her inconsideration of privacy and
security measures for the users of her chatbot.

One noble research was based on the pre-screening for female breast cancer [4], the authors here
used a variety of tools to help guide their research such as Chatbot Framework, YAML,
Character Based Word Segmentation, Logic Adapters. The SAC's assessment scores were
compared to the doctors' scores, and it achieved a high correlation accuracy. FBC doctors
evaluated based on scaling coefficient and weight factors. However, users giving too much
information might not give back any response due to lack of vocabulary.

Another research instance took the opportunity to work on Natural Language Processing based
Human Assistive Health Conversational Agent for Multi-Users [5]. The authors Jim Elliot
Christopher James; Mahima Saravanan; Deepa Beeta Thiyam; Prasath Alias Surendhar S used
Dialog flow application programming interface (API), which is powered by Google's Natural
Language Processing algorithm. The results of the paper demonstrate the development of a

5
health assistant system using Google's Dialog Flow API, which is a natural language processing
algorithm. Users can make health requests and receive relevant health suggestions and
recommendations through text messages.

In Health Chatbot: Design, Implementation, Acceptance and Usage Motivation [6], the authors
A. Softić, J. Husić, Aida Softić, Sabina Baraković made use of the Chat fuel platform to guide
their research. It allows users to interact with the chatbot in a conversational manner, helping
them identify their symptoms and decide whether or not to seek medical attention.

A simple yet effective system Chatbot for Healthcare System Using Artificial Intelligence [7],
The application uses a question and answers protocol where it consists of a login page. Query is
available in the database or displays similar answers. An expert answering page where experts
answer directly to the user’s question. Though, it failed to solve complex queries.

In Implementation of interactive healthcare advisor model using chatbot and visualisation [8],
Tae-Ho Hwang; JuHui Lee; Se-Min Hyun; KangYoon Lee made a system in which the user’s
biological signals could be supervised and forwarded to the chatbot for further observation.
Limited disease prediction and diagnosis using medical info. It requires artificial intelligence
systems.

Now let us look at the various projects already performed on this specific topic and gather key
takeaways from it such as tools used, results established and limitations of that paper.

6
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

1. An AI-Based IEEE Using NLP to User-friendly Takes more

Medical Innovations in interact chatbot time to give
Chatbot Intelligent directly with providing accurate
Model for Systems and users and give medical info results.
Infectious Applications responses. in various
Disease Conference languages,
Prediction (ASYU), 2023 offering
doctor
details,
hospital
locations,
and disease-
related
information,
facilitating
easy access.

7
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

2. Smart IEEE This chatbot The user's Only

Chatbot Transactions on mainly uses inquiries are inquiries
Architecture Systems, Man, NLP and NLU correctly from a closed
based NLP and Cybernetics: for its understood domain or
and Machine Systems, working. It by this questions that
learning for 2023 also uses ASR system, are defined in
health care i.e. Automatic which then the database
assistance. Speech provides are answered
Recognition. relevant by the
answers from chatbots.
the
knowledge
base in
response.

8
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

3. AI Based International This study has This The tool that

Healthcare Journal of Novel used the application enables
Chatbot using Research and following combines a marketers to
Natural Development techniques: language develop
Language (2023) NLP, Python. interpretation highly
Processing module, personalised
and Pattern sound-to- customer
Matching information interactions,
conversion, improve
and voice business
recognition response, and
software address client
issues.

9
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

4. Implementati International The primary Users can The training

on of an Journal of technology choose from data has a
Educational Innovative used for a list of major impact
Chatbot using Technology and building the courses in on the
Rasa Exploring chatbot is subjects like chatbot's
Framework Engineering Rasa. Content- Maths, accuracy.
(IJITEE) ISSN, based filtering Science, Insufficient
2022 is used for Computer training data
providing Science, and may lead to
personalised Aptitude. less accurate
recommendati Users can interpretation
ons take quizzes of user
related to the intents and
courses they entities.
have studied.

10
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

5. Natural Second In this study, This chatbot Answering a

Language International the health acts like an user's inquiry
Processing Conference on assistant informative requires extra
based Human Electronics and system was and time. To use
Assistive Sustainable developed conversation live chat,
Health Communication using Dialog al chatbot. there are fees.
Conversation Systems Flow This chatbot
al Agent for (ICESC), application provides
Multi-Users 2021 programming medical
interface (API) knowledge
which is a such as
Google's disease
Natural symptoms
language and
processing treatments.
powered
algorithm.

11
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

6. Online IEEE World Chatterbot The SAC's Users giving

Textual Conference on Framework assessment too much
Symptomatic Applied [9], scores were information
Assessment Intelligence and YAML, compared to might not
Chatbot Computing Character the doctors' give back any
Based on (AIC), 2021 Based Word scores, and it response due
Q&A Segmentation, achieved a to lack of
Weighted Logic high vocabulary.
Scoring for Adapters correlation
Female accuracy.
Breast FBC doctors
Cancer Pre- evaluated on
screening the basis of
scaling
coefficient
and weight
factors.

12
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

7. Health 20th This paper It allows Using

Chatbot: International presents the users to medical data
Design, Symposium health chatbot interact with to forecast
Implementati INFOTEH- application the chatbot in and diagnose
on, JAHORINA created on the a diseases is
Acceptance (INFOTEH), Chat fuel conversation limited. It
and Usage 2021 platform [10]. al manner, necessitates
Motivation helping them the use of
identify their artificial
symptoms intelligence
and decide technologies.
whether to
seek medical
attention.

13
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

8. College International The chatbot is Users can If the

Enquiry Journal of implemented find contact knowledge
Chatbot using Scientific using Rasa. information base is not
RASA Research in Conditional for the regularly
Computer Random Fields college, updated, it
Science, is a including the may not
Engineering and discriminative website and provide
Information model used for WhatsApp accurate
Technology predicting number. responses.
ISSN, sequences. The chatbot's
2021 effectiveness
heavily relies
on the
accuracy and
completeness
of the
information
stored in its
database.

9. Chatbot for International This paper The Complex

Healthcare Conference on used RDBMS, application queries may
System Using Reliability, n-gram, and uses a not be
Artificial Infocom cosine question and satisfied by
Intelligence Technologies, similarity. answers the database
and Furthermore, it protocol system.
Optimization utilised where it

14
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

(Trends and tokenization consists of a

Future and an expert login page.
Directions) system. Query is
(ICRITO), 2020 available in
the database
or displays
similar
answers. An
expert
answering
page where
experts
answer
directly to
the user’s
question.

15
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

10. Healthcare International N-gram Chatbot It takes more

Chatbot using Research Algorithm processes time to
Natural Journal of [12], TF-IDF user input respond to
Language Engineering and (Term with the user's
Processing Technology frequency- algorithms, question. Pay
(IRJET), 2020 inverse data clarifies some charges
frequency) symptoms to perform
[13], Cosine through live chat.
similarity questions,
algorithm categorises
disease as
minor/major,
suggests
nearby
doctors,
medications,
and
recovery-
oriented
food.

16
S. No. Paper Title Journal/ Tools/ Results Limitations
Conference Techniques/
(Year) Dataset

11. University IEEE AIML, Natural There are Complex

Chatbot using Conference on Language many queries may
Artificial Computer Processing, chatbots in not be
Intelligence Applications Pandorabots, English and satisfied by
Markup (ICCA) Pattern other the database
Language (2020) Matching, languages by system.
Response using
Generation different
algorithms
and models
but there is
little chatbot
using
Myanmar
language.

Fig. 2.1.1: Literature surveys consisting of tools, results and limitations observed

17
2.2 Key Gaps in the Literature

- Ethical issues: Data privacy, security, and prejudice are just a few of the ethical issues
surrounding healthcare chatbots. To employ chatbots in a safe and morally responsible
manner, further study is required to pinpoint and resolve these issues.

- User experience: It has not been given enough attention in study; instead, it is mostly
concerned with the technical features of chatbots rather than how people use and
perceive them. To create chatbots that are more efficient and easier to use, more study is
required to comprehend user demands and preferences.

- Integration issues: Integrating chatbots into the current healthcare systems presents a
variety of difficulties. To overcome these obstacles and improve the deployment of
chatbots, additional research is required.

- An absence of cooperation with medical experts: To further understand how

healthcare professionals see and use chatbots, more study is required. By ensuring that
chatbots are included into healthcare processes in a way that benefits both patients and
providers, this research will aid.

18
 CHAPTER 3: SYSTEM DEVELOPMENT

3.1 Requirements and Analysis

At this point, we will enumerate every prerequisite that the model needs. The different
technologies, libraries, and tools that are employed.

3.1.1 Python: The Basis for Readability and All-Inclusive Libraries

Python [14] is a popular, advanced, and versatile programming language. Its vast libraries and
ease of reading have contributed to its global appeal. It is the best option for numerous
applications, such as data analysis and web building.

- Readability: Python's syntax places a strong emphasis on the readability and simplicity of
code, which facilitates the writing and maintenance of code by developers. Large codebases
and collaborative projects will benefit from this functionality.

- Large Library: Python's vast library and framework ecosystem allow it to meet a wide
range of needs. These consist of libraries for machine learning (TensorFlow, PyTorch),
web development (Django, Flask), data analysis (NumPy, Pandas), and more. These
libraries' availability makes development easier and gives programmers powerful tools.

3.1.2 Streamlit: Making the Development of Web Applications Simpler

An open-source Python package called Streamlit [15] was made to make making dashboards
and web apps easier. With an emphasis on interaction and data presentation, it is excellent at
developing data-driven applications.

- Ease of Use: Developers can design web applications with little difficulty thanks to
Streamlit's user-friendly interface. It offers a straightforward script-based method that
expedites the design and prototyping process.

19
 - Data Visualization: Pre-built components are included in the collection to help with the
creation of dynamic and visually attractive visualisations. This makes it particularly
suitable for scenarios where it is imperative to communicate data insights through graphs
and charts.

3.1.3 The Natural Language Toolkit (NLTK): The Untapped Potential of Text Analysis

NLTK [16] is a powerful Python library designed specifically for natural language processing
(NLP). Along with other text analysis features, it provides an abundance of tools for sentiment
analysis, stemming, and tokenization.

- Text analysis capabilities: Because of its wide variety of text analysis skills, NLTK is
a popular choice for NLP work. Tokenization, which divides text into words, stemming,
which restores words to their most basic form, and other functionalities are available to
developers.

- Sentiment Analysis: We perform sentiment analysis through NLTK which will assist
us to ascertain if a certain statement depicts some positive or negative emotion. This
could assist in cases of applications like the capture of recipients input appropriately.

- Natural Language Processing: With NLTK's enhanced capabilities, programmers

may create computers that can comprehend, read, and produce text that is human-like.

3.1.4 Scikit Learn: For use of statistical modelling and machine learning models on it.

Scikit Learn [17] is a versatile python-based machine learning tool kit, having a lot of features
for different types of machine learning applications.

20
 - Simple and Consistent API: Clear and Consistent API has greatly contributed to Scikit-
Learn usage and good image among other machine learners. Its commitment to the
comfort of the practitioner is shown by the simple nature of the library’s API.

- Preparing data: It is a crucial phase in the machine learning process that attempts to
raise the quality of the dataset and prepare it for effective model training. Scikit-learn
offers a variety of tools and capabilities that handle many common preprocessing tasks,
making this process simple.

3.1.5 Pyfiglet:

A Python module known as pyfiglet [18] communicates with FIGlet, a program that generates
text banners in various typefaces using ASCII characters. The library provides Python
programmers with an easy way to incorporate FIGlet functionality into their scripts or programs,
giving users the capacity to create ASCII art text banners with a variety of font styles quickly
and easily.

- Integration with Scripts: Pyfiglet is a well-liked tool that developers use to add a
distinctive and visually appealing style to their command-line scripts or applications.
Applications using textual output can benefit greatly from this, as ASCII art can improve
the user experience.

- Font Customization: A variety of FIGlet fonts, each with a distinct style, are supported
by pyfiglet. By choosing various typefaces, users may alter how the created ASCII art
looks. Furthermore, the library offers developers freedom in font selection by enabling
the usage of bespoke FIGlet font files.

21
3.1.6 Pandas in Python:
Pandas [19] is an open-source data manipulation and analysis toolkit for Python that is both
robust and extensively utilised. Along with capabilities for data cleansing, aggregation, and
analysis, it offers data structures for effectively storing and managing massive datasets.

- DataFrame: The DataFrame, a two-dimensional, labelled data structure with columns

that may include various data kinds, is the main data structure in Pandas. It facilitates the
simple editing and analysis of organised data and is comparable to a spreadsheet or SQL
table.

- Series: A labelled array that is one dimension and is similar to a column in a

DataFrame. Dictionary entries, arrays, and lists can all be used to form series.

- Data Input/Output: CSV, Excel, SQL databases, JSON, and other file formats are
among the many file types that Pandas can read. It also offers ways to write data back
into these formats.

3.1.7 gTTS:
A Python package called "Google Text-to-Speech," or gTTS [20], makes it possible to convert
text to speech by using Google's Text-to-Speech API. You can turn any text into spoken words
with gTTS.

- Text-to-Speech Systems: Building voice assistants, virtual agents, or chatbots that

communicate with people orally is known as text-to-speech systems.

- Speech Synthesis: The text is processed by Google's API to create speech that sounds
natural. The API creates speech that sounds natural and human by using sophisticated
algorithms.

- Google Text-to-Speech API: gTTS uses the text that has been supplied as input when
sending a request to Google's Text-to-Speech API.

22
3.2 Project Design and Architecture
A thorough and careful approach is required for the design and architecture of a Python
healthcare chatbot, considering the complex needs and sensitivities of the healthcare sector. One
of the most important decisions in this process is selecting the system architecture. A scalable
and modular design, like microservices, can offer the flexibility required for future maintenance
and expansion.

Another important factor in guaranteeing the safe storage and effective retrieval of data
connected to healthcare is database design. Respecting privacy laws is essential to ensuring
patient data privacy and accuracy.

Natural language processing skills are the main focus of the chatbot's logic implementation,
which is frequently done by utilising well-known Python-based chatbot frameworks like
ChatterBot [21] or Rasa [22]. This makes it possible for the chatbot to comprehend and react to
user requests and questions.

Real-time access to patient records and data requires seamless integration with current
healthcare systems and databases. The chatbot's total usefulness is increased by this smooth
connection, which guarantees that it can respond with accuracy and current information.
Good communication and engagement are made easier by an intuitive and user-friendly
interface, which enhances the user experience.

The architecture incorporates security features like authorization procedures, secure

authentication methods, and encryption protocols. These safeguards are put in place to protect
private health information, guarantee legal compliance, and foster user confidence.

To provide a dependable and efficient healthcare chatbot solution, efficiency, security, and user
experience are essentially given top priority in the entire design and architecture. Through a
complete approach to these areas, the healthcare chatbot may be easily integrated into current
healthcare ecosystems, offering significant assistance to both users and healthcare professionals.

23
Fig 3.2.1: This is a rundown of the process of our project, the steps are implemented in
order to ensure a smooth implementation.

24
3.3 Data Preparation
Let us look at the statistics of our dataset which we are going to use in our model and the various
variables involved in it.

Primary dataset is comprised of all kinds of diseases with multiple permutations and
combinations of possible and likely symptoms.

TOTAL NUMBER OF DISEASES RANGE OF NUMBER OF OBSERVED

(INCLUDING ALL POSSIBLE SYMPTOMS IN EACH INSTANCE
INSTANCES)

4920 3-18

Fig 3.3.1: Description of total instances and symptoms available in the dataset

We have further information about each of the diseases, the symptoms corresponding to them
and the severity of the exposed symptoms. We have also gathered the possible precautions a
patient can avail when they are exposed to a specific disease.

UNIQUE UNIQUE RANGE OF RANGE OF

NUMBER OF NUMBER OF SEVERITY OF NUMBER OF
DISEASES SYMPTOMS SYMPTOMS PRECAUTIONS
PER EACH INST.

41 132 1-7 3-4

Fig 3.3.2: Description of unique number of diseases, symptoms and precautions,

severity of symptoms is also a key to alarm the patient

The most important step in any algorithm to separate the dataset between training and testing
phase. If the training accuracy is ideal, only then the testing phase can be considered fit for trial
and hence the algorithm can be successfully implemented.

25
 PHASE OF THE DATASET NUMBER OF INSTANCES

PRIMARY 4920

TRAINING 4879

TESTING 41

Fig 3.3.3: Description of instances available in various phases of the dataset.

After splitting the dataset, the symptoms are then binary encoded as it is easier to work with
binary numbers.

To guarantee the efficacy and precision of a healthcare chatbot, data preparation is a rigorous
and multi-step procedure. Clearly describing the chatbot's precise use cases—such as symptom
assessment, prescription writing, or patient education—is the first stage in the process. This
stage makes sure that the model created and the data gathered are appropriate for the intended
uses.

Clinical records, medical literature, and patient encounters are among the sources from which
pertinent healthcare data is acquired. To guarantee excellent quality and consistency, the
gathered data is thoroughly cleansed. To prevent redundancy, this entails eliminating duplicates,
standardising formats for consistency, improving accuracy by fixing mistakes, and managing
missing information to preserve dataset integrity.

First and first, it is essential to identify particular use cases, which might range from evaluating
symptoms to writing prescriptions. After relevant healthcare data is gathered from sources such
as clinical records and medical literature, a thorough data cleansing procedure is performed. To
maintain data quality and consistency, this entails addressing problems including duplication,
standardising formats, fixing mistakes, and managing missing values.

26
The cleaned data is labelled appropriately for supervised learning. This is an important stage
because it lets the model identify links and patterns in the data. Depending on the use case of
the chatbot, annotations may involve labelling medical processes, diagnoses, or symptoms.
After the text has been analysed, feature extraction techniques are used to turn it into numerical
features. This conversion converts text data into an algorithm-readable format, which is essential
for training machine learning models.

The processed text input is subsequently converted into numerical features by feature extraction,
which makes it easier to train machine learning models. To ensure that the model is trained on
one subset and validated and tested on independent subsets to assess its generalizability, the
dataset is thereafter divided into training, validation, and testing sets.
The training, validation, and testing sets are the three subsets that make up the dataset. This
section is essential for evaluating the generalizability and performance of the model. The model
is trained on the training set; it is fine-tuned and protected from overfitting by the validation set;
and its ultimate performance is assessed on the testing set.

Models are trained to comprehend user questions and provide intelligent responses using the
provided dataset. The model gains knowledge from the training set and is refined throughout
training in response to input from the validation set. The last assessment on the testing set
verifies that the model functions effectively on untested data, demonstrating its resilience and
dependability.

These procedures must be carefully followed to supply the healthcare chatbot with a high-quality
dataset that allows it to respond intelligently and accurately to a range of healthcare-related
questions. This careful planning guarantees that the chatbot will be able to assist patients and
healthcare providers in an efficient manner.

The following table describes how a single disease can have different symptoms and negates the
point that only certain symptoms can be directly associated to a particular disease.

27
Fig 3.3.4: The above figure of our dataset shows various diseases and their corresponding
symptoms. This helps our model to match the disease with likely symptoms.

28
To initiate data preparation, we convert the symptoms into binary form by binary encoding.

Fig 3.3.5: The diseases are then binary encoded which makes it easier to train
the model as working on binary digits is way easier.

29
3.4 Implementation
Using Natural Language Processing (NLP) techniques to implement a healthcare chatbot
requires several important processes, from selecting the appropriate tools and libraries to
creating and launching the chatbot. This is a thorough overview of the implementation
procedure:

● Select Libraries and Frameworks:

Choosing Python frameworks and modules that meet the needs of the project.

Fig 3.4.1: List of some of the pre included libraries and packages in python which we have
made use of.

30
● Preprocessing:
While preparing text data for NLP analysis, tokenization, stop word removal, and
lemmatization or stemming are essential processes. The text is divided into discrete
words, or tokens, using tokenization. By eliminating frequent words like "and" or "the,"
which typically do not add much sense, stop words are eliminated. Words are reduced to
their base forms by lemmatization, and to their root forms through stemming. Through
the reduction of noise and concentration on significant material, these procedures
guarantee that the text is in a consistent and streamlined structure that is appropriate for
efficient NLP analysis.

Fig 3.4.2: The dataset is divided into two categories: training data and
testing data to help in better performance.

31
● Model Selection:
Based on our case, we select a suitable NLP model. Conventional machine learning
models or rule-based models can be used for easier jobs.

Fig 3.4.3: Configuration file of our model which lists down various frameworks
and their versions to run our code.

● Training the Model:

Utilising the training set, train the model, then adjust its performance using the validation
set. Using healthcare-specific data to refine the pre-trained models for deep learning
models.

Fig 3.4.4: The data is then trained where the disease is matched with the symptoms.

32
● Integration with Chatbot Framework:
Construct a connection between the chatbot framework and the NLP components.
Through this connection, the chatbot's comprehension and suitable response to user
inputs are guaranteed.

Fig 3.4.5: The user can enter their symptoms in the chatbot.

● User Interface Development:

Create an intuitive user interface so that people can communicate with the chatbot. This
might be an app for a web interface, or a connection with already-in-use communication
systems.

Fig 3.4.6: Interface of the chatbot.

33
● Testing:
Introduce the chatbot to a comprehensive testing session. Verify that it delivers accurate
and contextually relevant information by testing its replies in a range of healthcare
contexts. Conduct user acceptability, integration, and unit testing.

Fig 3.4.7: Interaction between the chatbot and the user.

34
● Monitoring and Maintenance:
Using reliable monitoring solutions is essential for keeping track of user interactions and
evaluating chatbot effectiveness when it comes to monitoring and maintenance. These
technologies make it easier to see possible problems, which enables prompt intervention
and optimisation. Simultaneously, an organised maintenance routine is essential. The
continuous efficacy and relevancy of the chatbot are contingent upon regular updates,
problem fixes, and upgrades. In this process, user input plays an important role as a
compass, helping to prioritise changes and enhancements. Through adherence to a
proactive approach that includes regular monitoring and a methodical maintenance
schedule, companies may cultivate long-term user engagement and happiness,
strengthening the chatbot's usefulness and durability.

● Documentation:
Developers, administrators, and end users are guided towards a full grasp of our chatbot
system with this extensive documentation. Customised for every user group, the
documentation describes the features of the chatbot in great depth and offers a thorough
explanation of each one's capabilities. Users are provided with thorough instructions on
how to use the chatbot to expedite their activities, ranging from simple requests to
complex capabilities. Furthermore, troubleshooting instructions are carefully selected
and provide enlightening answers to frequent problems that could occur during use.
Through summarising the chatbot's capabilities and addressing potential issues, this
handbook functions as a vital resource for optimising efficiency and expertise for users
of all skill levels.

35
 ● Working with the dataset:

Fig.3.4.8 Symptom precaution.

The figure 3.4.8 describes various diseases and the precautions that a patient can take when they
contract such a disease. Multiple precautions are provided corresponding to a single disease so
the patient can act more cautiously.

36
Following is the snippet of various symptoms, these are associated to a severity number on a
scale of 1-7, 1 being least severe to 7 being most severe.

Fig 3.4.9 Symptoms severity ranging from 1-7.

We also have associated another sub-dataset that describes every disease in a summarised
and to the pinpoint description. This helps the patient to better know about what they are
dealing with without having to research about their disease in a detailed manner.

37
 Fig.3.4.10 Disease description dataset.

3.5 Key Challenges

● Managing Unstructured Data:
Data from clinical notes, medical records, and patient-reported symptoms are just a few
examples of the varied and sometimes unstructured nature of healthcare information. It
is quite difficult to develop NLP models that can successfully extract useful information
from such unstructured input.

● Guaranteeing Precision and Dependability:

Since errors can have major repercussions, healthcare chatbots must respond and provide
accurate information. It is a constant challenge to ensure the reliability of NLP models,
particularly in critical medical situations.

● Limited Labelled Data:

It will be challenging for the chatbot to respond to a question that is not covered by the
data it has been given because of the small amount of information it has been given.
There will be few responses as a result.

● Cultural and language variances:

When it comes to how users explain their health issues or express symptoms, healthcare
chatbots must take into consideration cultural and language variances. One ongoing
problem is adapting the model to different linguistic styles and cultural quirks.

38
 CHAPTER 4: TESTING
4.1 Testing Strategy
It is vital to do thorough testing on a healthcare chatbot that employs Natural Language
Processing (NLP) methods to guarantee precise, dependable, and secure functionality. An
extensive testing plan for the report is provided below:

● User Acceptance Testing (UAT): Involve actual users in conversation with the
chatbot in a variety of healthcare contexts, gathering input on responding and usability.

● Performance Testing: Time responses, gauge scalability in different scenarios, and

test under extreme load situations to gauge performance.
Response Time: Calculate how long it takes to create responses based on user input.
Throughput: Evaluate the chatbot's throughput by counting how many queries it can
process in a second.
Scalability: Evaluate the chatbot's capacity to function as the load increases.

● Usability Testing: Evaluate the user interface's intuitiveness while taking user
comments and ease of use into account.

● Integration Testing: The goal of integration tests is to make sure that the various
components of the chatbot function together as planned.
Database interaction: To store and retrieve user data, make sure there is a smooth
interaction with the database.

● Functional Testing: Functional testing confirms that the chatbot accurately carries out
tasks linked to healthcare.
Symptom Checking: Ensure if the chatbot can respond to questions about symptoms.
Medical Advice: Ensure that the chatbot offers accurate and suitable medical advice.

39
4.2 Test Cases and Outcomes

Assessing a model's performance is essential to establishing its effectiveness, especially in high-

stakes environments like the healthcare industry. This review procedure is like software
development in many aspects, as test cases are essential. Test cases entail feeding a model a lot
of inputs and examining the outputs that come out. This methodology guarantees the model's
ability to manage an extensive array of situations, showcasing its adaptability and resilience.

A model intended for diagnostic use, for example, might be evaluated in the healthcare area
utilising a variety of patient profiles and symptoms. Using a variety of data points, we may
evaluate the model's diagnostic accuracy for various circumstances. This comprehensive
analysis is essential since it confirms that the model's predictions hold true in a variety of
possible real-world scenarios. Discrepancies between the model's outputs and predicted results
might be used to identify areas for model improvement.

Determining and resolving the model's biases and shortcomings requires extensive testing with
a variety of inputs. This might entail making sure that the model works equally effectively for
all demographics, such as age groups, genders, and ethnicities, in a healthcare context in order
to prevent biases that can result in uneven treatment. Any biases found may be fixed by rigorous
testing, improving the model's overall fairness and accuracy.

To sum up, test case performance evaluation is essential for confirming the precision of
predictive models, especially in vital domains such as healthcare. It guarantees the robustness
and reliability of the model and aids in locating biases and potential improvement areas. We can
improve the performance of models and make sure they deliver reliable and accurate predictions
in real-world applications by subjecting them to rigorous testing using a variety of inputs.
Building trust in the model's predictions and, eventually, enhancing decision-making in real-
world situations depend heavily on this procedure.

40
 - User experiencing stomach ache:

Fig 4.2.1: The user is interacting with the chatbot and replying to the queries.

Outcome: User may have Gerd.

41
- User experiencing weight loss

Fig 4.2.2: Feeding the symptoms to the chatbot and getting the likely disease.

Outcome: User may have jaundice

42
 CHAPTER 5: RESULTS AND EVALUATION

5.1 Results
User authentication procedures provide safe access to patient-specific data, and thorough
security testing verifies the chatbot's resistance to typical security risks. The chatbot responds to
unclear requests with clarity or broad information, and it handles errors with efficiency. The
chatbot's usefulness in urgent medical conditions is demonstrated by its smooth emergency
response capabilities.

The chatbot's ability to estimate symptom severity is precise and sophisticated, demonstrating
its mastery of medical context awareness. The system has processes in place for quickly
elevating serious problems, and it responds to user feedback with effectiveness. The chatbot
exhibits scalability even under heavy loads, maintaining constant performance across many
browsers. The chatbot complies with security and privacy regulations, demonstrating its
regulatory compliance.

The chatbot's medical correctness and suitability for healthcare settings are further supported by
positive comments from cooperative testing with medical experts. All these outcomes support
the chatbot's effectiveness, privacy compliance, and user-centred design in providing
trustworthy and safe medical information. It is advised that it be continuously monitored and
adjusted to new standards to maintain its capabilities.

43
Fig. 5.1.1 Result of the working model when data is fed to it

44
5.2 Comparison with Existing Solutions
The use of technology has opened the door for creative ways to improve patient involvement
and expedite communication in the quickly changing healthcare environment. The use of
Natural Language Processing (NLP) methods in healthcare chatbots is one such innovative use.
To give a thorough knowledge of the advantages and disadvantages of the current solutions on
the market, this comparative study attempts to assess and compare them.

● Accuracy of Symptom Identification:

In contrast to previous alternatives, the healthcare chatbot excels in symptom
identification accuracy. Its powerful NLP algorithms enable nuanced interpretation of
user input, considerably boosting the precision of sickness identification. This
heightened accuracy separates the chatbot, establishing it as a more dependable and
advanced tool for users seeking precise health evaluations through natural and intuitive
interactions.

● Context-Aware Prescription Generation:

Distinguished from existing methods, the created chatbot excels in identifying diseases
from symptoms. Its major feature is in individualised drug suggestions, considering the
user's medical history and reported symptoms. This complete approach not only sets the
chatbot distinct in correct diagnosis but also positions it as a user-centric healthcare
assistant, delivering individualised recommendations for a more holistic healthcare
experience.

● Compliance with Security and Privacy:

Distinguished by rigorous privacy and security features, the chatbot assures compliance
with demanding healthcare standards like HIPAA. This remarkable commitment to
securing private medical data sets it apart from several existing solutions that may lack
comparable security safeguards, demonstrating the chatbot's focus to ensuring the
confidentiality and integrity of sensitive healthcare information.

45
 ● User-Friendly Interface and Responsiveness:
In contrast to others, the chatbot has a highly adaptable and user-friendly interface
suitable with numerous devices. Its excellent responsiveness offers a smooth user
experience, resulting in heightened user engagement. This versatility, along with a
flexible design, distinguishes the chatbot, making it an accessible and user-centric
solution that caters to varied user preferences and devices.

● Complete Guidance for Prescription Management and Disease Identification:

Surpassing certain existing solutions, the chatbot excels in complete sickness diagnosis and
prescription guidance. Its comprehensive technique provides users with smart advice,
evaluating a wide diversity of symptoms, boosting their comprehension of health
conditions. This complex diagnostic capacity presents the chatbot as a more intelligent and
dependable source for people seeking thorough and accurate health information.

Every healthcare chatbot solution that has been described has its own advantages and
disadvantages and aims to meet various healthcare business demands. The organisation's
objectives, financial limits, and desired degree of customisation should all be taken into
consideration while selecting a particular solution. The capabilities and effectiveness of
healthcare chatbots will probably continue to grow as the area develops thanks to continuous
research and advancements in NLP.

46
 CHAPTER 6: CONCLUSION AND FUTURE
SCOPE

6.1 Conclusion
The healthcare chatbot, which uses precise Natural Language Processing (NLP) technology to
expedite preliminary health evaluations, is a major improvement in virtual healthcare support.
Through conversational engagement and eliciting symptoms, the chatbot is able to provide
precise illness diagnosis and customised medication recommendations.

This chatbot stands out because of its strong natural language processing (NLP) algorithms,
which are further improved by a prescription tool that considers context to provide tailored
suggestions. By giving personalised guidance, this not only increases the accuracy of diagnoses
but also improves the user experience overall.

A seamless user experience is ensured by the chatbot's adaptable design, which enables seamless
interaction and adjustment to various user demands and preferences. Furthermore, cooperative
testing with medical professionals raises the chatbot's legitimacy and establishes it as a reliable
health advisor.

Frequent upgrades are necessary for it to remain current and keep becoming better. These
improvements not only show the system's dedication to improving virtual healthcare assistance, but
they also guarantee privacy standards are followed and a user-centric design approach is maintained.
All things considered, the healthcare chatbot is a promising advancement in virtual healthcare,
providing precise diagnosis, tailored advice, and a smooth user experience.

47
6.2 Future Scope
The future scope of the healthcare chatbot employing NLP techniques is promising, with
various paths for extension and enhancement:

● Web Server Deployment with Mobile Application:

The initial focus should be on installing the chatbot on a web server to facilitate
accessibility. Additionally, establishing a specialised mobile application will increase
the reach, allowing consumers to receive healthcare information effortlessly from their
cell phones.

● Integration of Facial Recognition:

The incorporation of facial recognition technology provides an interesting future
opportunity. By implementing face recognition, the chatbot may possibly examine
apparent symptoms or emotions, adding another dimension to its diagnostic skills.

● Advanced Diagnostic Capabilities:

The function of the chatbot as a trustworthy first health counsellor will be further
cemented by enhancing its diagnostic abilities with more sophisticated medical
knowledge and keeping up with the most recent medical research.
Expanding the chatbot's diagnostic capabilities by incorporating more advanced medical
knowledge and staying updated with the latest medical research will further solidify its
position as a reliable preliminary health advisor.

● Personalized Health Recommendations:

Moving beyond generic prescription suggestions, the chatbot could evolve to provide
more personalised health recommendations based on user profiles, medical histories, and
lifestyle factors, augmenting the user's overall healthcare experience.

48
● Accessibility Features:
Enhancing accessibility features, such making sure the interface is easy to use for people
with impairments and being compatible with screen readers, would make the healthcare
system more inclusive.

● User Engagement Techniques:

Personalised health advice, frequent health check-ins, and push alerts for health
reminders are a few examples of tactics that may be used to improve user engagement
and make healthcare more proactive and focused on the needs of the individual.

● Health Tracking and Notifications:

Real-time monitoring features may be added to chatbots by using natural language
processing (NLP) to analyse data from wearables and other health sensors. Changes in
health parameters can trigger automated warnings and reminders that guarantee prompt
action and preventive care.

49
 REFERENCES
1) https://www.nih.gov/ : https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6166510/

2) “Healthcare Chatbot using Artificial Intelligence,” Ijraset.com, 2022.

https://www.ijraset.com/research-paper/healthcare-chatbot-using-artificial-intelligence

3) G. Kurup and S. D. Shetty, “AI Conversational Chatbot for Primary Healthcare

Diagnosis Using Natural Language Processing and Deep Learning,” ResearchGate,
2022.
https://www.researchgate.net/publication/357162231_AI_Conversational_Chatbot_for
_Primary_Healthcare_Diagnosis_Using_Natural_Language_Processing_and_Deep_Le
arning.

4) J.-H. Chen et al., “Online Textual Symptomatic Assessment Chatbot Based on Q&A
Weighted Scoring for Female Breast Cancer Pre-screening,” Applied sciences, vol. 11,
no. 11, pp. 5079–5079, May 2021, doi: https://doi.org/10.3390/app11115079.

5) Jim Elliot Christopher James et al., “Natural Language Processing based Human
Assistive Health Conversational Agent for Multi-Users,” 2021 Second International
Conference on Electronics and Sustainable Communication Systems (ICESC), Aug.
2021, doi: https://doi.org/10.1109/icesc51422.2021.9532913.

6) A. Softic, Jasmina Barakovic Husic, A. Softic, and S. Barakovic, “Health Chatbot:

Design, Implementation, Acceptance and Usage Motivation,” Mar. 2021, doi:
https://doi.org/10.1109/infoteh51037.2021.9400693.

7) Lekha Athota, Vinod Kumar Shukla, N. Pandey, and A. Rana, “Chatbot for Healthcare
System Using Artificial Intelligence,” Jun. 2020, doi:
https://doi.org/10.1109/icrito48877.2020.9197833.

50
8) T.-H. Hwang, J. Lee, S.-M. Hyun, and K. Lee, “Implementation of interactive healthcare
advisor model using chatbot and visualisation,” Oct. 2020, doi:
https://doi.org/10.1109/ictc49870.2020.9289621.

9) “AN APPROACH TO AI BASED HEALTHCARE CHATBOT SYSTEM BY USING

NLP.” https://ijcrt.org/papers/IJCRT2208340.pdf

10) “A STATISTICAL INTERPRETATION OF TERM SPECIFICITY AND ITS

APPLICATION IN RETRIEVAL | Emerald Insight,” Journal of Documentation, vol.
28, no. 1, pp. 11–21, 2024, doi: https://doi.org/10.1108//jd.

11) “AI Based Healthcare Chatbot using Natural Language Processing and Pattern
Matching,” © 2023 IJNRD , vol. 8, no. 5, p. 793, 2023.
https://www.ijnrd.org/papers/IJNRD2305598.pdf

12) C. Bartneck et al., “Human-Robot Interaction An Introduction.” Available:

https://www.human-robot-interaction.org/wp-content/uploads/2019/08/human-robot-
interaction-an-introduction-web.pdf

13) J. Chang, J. Park, and J. Park, “Using an Artificial Intelligence Chatbot in Scientific
Inquiry: Focusing on a Guided-Inquiry Activity Using Inquiry Bot,” Asia-Pacific science
education, vol. 9, no. 1, pp. 44–74, Jun. 2023, doi: https://doi.org/10.1163/23641177-
bja10062.

14) Python Language: https://www.python.org/

15) Streamlit: https://streamlit.io/

16) NLTK: https://www.nltk.org/

51
17) Scikit-learn: https://scikit-learn.org/stable/

18) Pyfiglet: https://pypi.org/project/pyfiglet/

19) Pandas: https://pandas.pydata.org/

20) gTTS: https://pypi.org/project/gTTS/

21) Chatterbot: https://chatterbot.readthedocs.io/en/stable/

22) RASA: https://rasa.com/

52
53
54