EasyChair Preprint

№ 5260

AI Based Healthcare Chatbot System

A Kumaresan Angappan, N Jazeem Khan, M Sharan and

S Fazil Ahamed

EasyChair preprints are intended for rapid

dissemination of research results and are
integrated with the rest of EasyChair.

March 31, 2021

 AI Based Healthcare Chatbot System
A Kumaresan Angappan N Jazeem Khan M Sharan S Fazil Ahamed
Department of Information Department of Information Department of Information Department of Information
Technology Technology Technology Technology
Hindustan Institute of Technology Hindustan Institute of Hindustan Institute of Hindustan Institute of
and Science, Chennai, India Technology and Science, Technology and Science, Technology and Science,
[email protected] Chennai, India Chennai, India Chennai, India
[email protected] [email protected] [email protected]

ABSTRACT:
Medical field has changed a lot in the past known as conversational agents, interactive
decade. Research says 60% of patients visit agents, virtual agents, virtual humans, or
doctor for common health issues and 80% of virtual assistants, are artificial intelligence
the diseases could be cured using homemade programs designed to simulate human
remedies. In this pandemic we could decrease conversation via text or speech. [5,6]
the physical contact by the usage of medical
chatbots which will provide herbal methods Interactive chatbot applications are the latest
to cure the disease in home itself. By using a inventions of modern era. The healthcare
healthcare chatbot people can avoid sector is closely associated with human
unnecessary visit to clinics and hospitals. interaction, and it seems that conversational
Especially in remote areas, it is becoming AI applications like chatbots are more
more difficult to consult a medical specialist prevalent. A chatbot should respond in a way
when there is an emergency situation. Thus, that the user should feel like they are
an effectively designed and deployed chatbot
conversing with a real person. The chatbot
can help patients who are living in remote
response according to the clear dataset and
areas by providing preventive measures and
home remedies. sustainable backend logic for the outcome
generation. A medical chatbot facilitates the
job of a healthcare provider and helps to
improve their performance by interacting
Keywords: with users in a human-like way. Chatbots in
Artificial Intelligence Mark-up Language health care may have the potential to provide
(AIML), Herbal, Healthcare-chatbot, patients with access to immediate medical
Interactive agent, Natural Language information, recommend diagnoses at the
Processing (NLP), Pattern Matching. first sign of illness, or connect patients with
suitable health care providers (HCPs) across
their community.[3]
1.Introduction:
Chatbots do not get tired, fatigued, or sick,
The rise of web and mobile applications have and they do not need to sleep; they are cost-
paved way for many interventions in the field effective to operate and can run 24 hours a
of medicine and healthcare. Chatbots, also day, which is especially useful for patients
who may have medical concerns outside of contacting health professionals or providing
their doctor’s operating hours. information about immediate preventive
measures. The main purpose of this chatbot is
2.Related Works: to make the most of that disease - Covid 19.
The criteria for creating chatbot for medical Ghare Shifa et al. [12] developed a
applications should be efficient. Gopi dialogue based on Artificial Intelligence to
Battineni, et al. [13] designed complex AI reduce the need for a physician and reduce
medical discussions for users, especially in medical costs. Artificial Intelligence (AI) as
times of unknown illnesses such as nCOV- the building block of learning a new complex
19. It is, therefore, clear that if well-designed life and working for an ongoing circle. The
and well-used conversation can help patients user will have a text chat with the healthcare
living in remote areas by promoting immune chatbot and then get to identify the primary
systems, regenerating viruses, and reducing disease and the user can also access their chat
the psychological damage caused by isolation records within their features collected in the
and fear. This study reveals the complexity of database. They have used 3.7 python version
artificial intelligence (AI) discussion for the
in jupyter notebook. NLP is used to detect
and analyse human language. BLU is a subset
of the larger NLP image. It processes raw
data which is considered a chatbot brain,
which cleans it and provides appropriate
actions. The proposed system is used only to
detect disease but does not provide any
remedies.
Aishwarya Kedar et al. [4] aimed to make a
chatbot system with the help of artificial
intelligence and machine learning, our
system is based on hospital management.
Each user must log in to the system to use it,
we provide a chatbot of hospitals that will
serve as appointment bookings, details of
Fig 1: Working Flowchart
surgeons, presence of doctors etc. we make a
purpose of diagnosing and recommending smart plan that will make the user get general
immediate action when patients are exposed hospital information with finger tips. These
to nCOV-19. In addition, introducing a real bots interact with potential patients who visit
assistant can also measure the severity of the the site, help them find specialists, book their
infection and contact a registered physician appointments, and get them the right
when the symptoms worsen. With AIML treatment. By asking a series of questions it
integration in our chatbot, it can detect helps patients to direct what they want. The
patterns from user messages and provide resulting chatbot focuses on operating only in
direct and logical responses. Once it has all the hospital area.
the necessary information, it detects the acute
percentage of the virus that the user Dinesh Kalla and Fnu Samaah [9]
encounters and acts appropriately, by developed an application that uses Artificial
Intelligence and can help diagnose a variety 3.Proposed Work:
of diseases and provide the necessary
The proposed chatbot program aids the work
information about a patient's illness. It uses
of the health care provider and helps to
techniques from Natural Language
improve their performance by
Processing, which works in human language
communicating with users in a human-like
to provide feedback. This app answers the
interaction rather than the system manner.
same questions that medical professionals
The human-like chatbot mainly works based
may have answered, their purpose in
on AIML (Artificial Intelligence Mark-up
answering objectives is to take input from the
Language) code. AIML is an XML
patient or user, process it using specific
(Extensible mark-up language) based mark-
methods and provide results to the user
up language meant to create artificial
according to their symptoms. In the first stage
intelligent applications. AIML makes it
they process the input text by translating the
possible to create human interfaces while
whole sentence into lowercase letters so that
keeping the implementation simple to
the algorithm does not contain the same type
program, easy to understand and highly
of words from uppercase letters differing
maintainable.
from the same word, from Chatbot for
Medical Treatment using NLTK Lib. Since The Spring Tool Suite is used to run the
the output has a very low cosine similarity, AIML file, which is a JAVA based platform
and the response may or may not be exactly highly useful for developing web
the same. applications. In this chatbot, the user types
the question in the chat box, the chatbot
Ashwini Shangrapawar [3] et al
compares the question pattern with the AIML
designed a chatbot is to give service to the
code. Then the process of pattern matching is
people who are suffering from body problem
carried out to make the chatbot respond to
or disease by suggesting them medicines
respective questions. Pattern matching is the
regarding their problems. To give them 24x7
process of checking whether a particular
availability, we use our hardware boot as a
sequence of letters / data is present in a given
server for client using programming. This
data. The main feature of this chatbot is the
medical chatbot provides medical assistance
user gets an idea about herbal remedies which
to the patients for some of the general
are more effective, healthy and safe way of
diseases like fever, cold, typhoid, malaria,
treating illnesses. Generally, the main
jaundice etc. Here they applied input speech
objective of any Chatbot system is that it
signal through the mic and it compared with
should behave in a human like way. That is,
the database stored in the raspberry pi and
the user must feel like they are actually
answer from the database. If problem
chatting with a healthcare provider.
statement or disease and medicine related
information is not available in the database Hence, we also aim to make this Chatbot
then, it will answer from the Google server. more user friendly.
The drawback of this proposed system is that
it is a hardware-based system which should
be carried from place to place.
 Various types of AIML tags are used in the
process of creating the chatbot. Some of the
important tags used in the AIML are Pattern
tag, Template tag and category tag. The
<pattern> is used to match the users input
text with the pre-stored data. The <template>
is used to store the data for which the bot will
respond according to the user’s input. The
<category> is used to define the unit of
knowledge in our AIML bot.

4.1. Algorithm:
Fig 2: System Process Input: Natural Language Query from user
4.Simulation and Analysis: Output: Requested data from CSV file or
1.Datasets: The dataset is based on the CSV AIML tags based on input in form of Natural
files. A CSV is a comma separated values file Language.
which allows data to be saved in a table like Pseudocode: User Interaction.
format. It can be saved in n excel sheet with
a .csv extension. This dataset contains the 1.While user has a question.
details of all the common disease that are
2.User enters the query.
already analyzed and shortlisted, also their
corresponding herbal remedies and methods 3.detected and extracted words.
for those particular disease and its related
symptoms has been pre-saved in the CSV 4.for each keyword k in query.
file. 5. Lookup k in CSV data fields.
2.Pattern Matching: It is the process of
6. Lookup k in <pattern> of AIML.
checking whether a sequence of
data/keywords are present in the given 7. return the CSV data value v.
dataset. Pattern Matching reads through text
in dataset. The name of the disease and 8. end for.
symptoms are represented as keywords are 9. concatenate v with predefined reply sting.
already stored within the pattern tag
<pattern> of the AIML. By using AIML tags 10. end while.
Pattern-matching is simple and more
accurate.
Symptoms are presented as keywords and are
stored within the pattern tag <pattern>of the
AIML. By using AIML tags Pattern-
Matching is simple and more accurate.
4.2. Model Accuracy Graph: above graph shows the various characteristics
of a chatbot system such as Accuracy,
Efficiency, Pattern matching and the
solutions given by the chatbots. Therefore, by
analysing the abilities of the chatbots, the
graphs are drawn.
4.3. Sample output:
The sample output of our proposed system
shows the conversation between the user and
the chatbot system.

-Dr. Bot
Fig 3: Model Accuracy Graph
The above graph represents the Model
accuracy. It is a graphical representation of
factors such as Accuracy, efficiency of the
proposed system. The pattern Matching
ability and the solutions given by the system
is also shown in the graph.

Fig.5: sample output

5.Conclusion and Future work:

Healthcare chatbots are the future of medical
field as it aids in reducing the amount of
physical contact between patient and the
-Dr. Bot - Other Chatbots doctor in the day to day growing population.
Fig.4: comparison graph Our chatbot (Dr.bot) uses natural language
processing to interact with the user. Dr.bot
The above graph is a comparison of our uses pattern matching to recognise the user
proposed system along with similar other input and provide a suitable response from
chatbots. The blue line represents the abilities the provided dataset.
of other chatbots. The orange line represents
the abilities of our proposed system. The
The proposed system will include a brief of-bots-in-healthcare-4d42c5539547[Web
summary of herbal medicines, their uses and
Cite]
suitable home remedies that can be used to
treat and cure most of the common diseases. [7] Cameron, G., Cameron, D., Megaw, G.,
Bond, R., Mulvenna, M., Neill, S. O,
6.References:
McTear, M. (2018). Best Practices for
[1] Adam Palanica, “Physicians Perceptions Designing Chatbots in Mental Healthcare – A
of Chatbots in Health Care: Cross-Sectional
Case Study on iHelpr. Proceedings of British
Web-Based-Survey”,JMIR
Publications:Published on 05.04.19 in Vol HCI 2018
21, No 4 (2019): April. [8] Chin-Yuan Huang, Ming-Chin Yang,
Chin-Yu Huang, “A Chatbot-supported
[2] Amato F, Marrone S, Moscato V,
Piantadosi G, Picariello A, Sansone C. CEUR Smart Wireless Interactive Healthcare
Workshop Proceedings. 2017. Chatbots Meet System for Weight Control and Health
eHealth: Automatizing Promotion”, proceeding of the IEEE, April-
Healthcare URL: http://ceur-ws.org/Vol- 2018.
1982/paper6.pdf [accessed 2019-02-26]
[9] Dinesh Kalla, Fnu Samaah,” Chatbot for
[Web Cite]
Medical Treatment using NLTK Lib”
[3] Ashwini Shangrapawar, Ankita
IOSR Journal of Computer Engineering
Ravekar,” Artificial Intelligence based
Healthcare Chatbot System”, IRJET Vol:07, IOSR-JCE, Volume 22, Issue 1,
Issue-02:Feb-2020. Feb -2020.
[10] Divya, Indumathi, Ishwarya,
[4] Aishwarya Kedar, Jyoti
Priyasankari, “A SelfDiagnosis Medical
Dahale,Khushboo Patel,” Chatbot System for
Chatbot Using Artificial Intelligence”,
Healthcare using Artificial Intelligence”,
proceeding MAT Journal, October-2017.
September 2020 IJSDR | Volume 5 Issue
[11] Flora Amato, Stefano Marrone,
[5] Brooke S. Chatbots Magazine. How
“Chatbots meet eHealth: automat zing
chatbots will shape the future of
healthcare”, proceeding of diet, May-2018
healthcare URL: https://chatbotsmagazine.
[12] Ghare Shifa,Shaikh Sabreen,” Self-
com/how-chatbots-will-shape-the-future-of-
Diagnosis Medical Chat-Bot Using Artificial
healthcare-
Intelligence”,Easychair Preprint No.2736-
fa8e30cebb1c?gi=7238e40fd943 [accessed
2020.
2019-02-25] [Web Cite]
[13] Gopi Battineni, Jadhav, K.P,” AI
[6]Cade J. Medium. No appointment
Chatbot Design during an Epidemic like the
necessary — The rise of bots in
Novel Coronavirus”, Special Issue:Feature
healthcare URL: https://medium.com/gene-
Papers in Artificial Intelligence in
global/no-appointment-necessary-the-rise-
Medicine),2020.