Scribd
Upload
5 views

B.tech Project 1

Uploaded by

fgshwjwj177.8
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

B.tech Project 1

Uploaded by

fgshwjwj177.8
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

B.

Tech Project I (2024-25)


Project Supervisor: Prof. Vinod Kumar

Medical Chatbot using


Generative AI
Dhruv Arora Gagandeep Singh
2K21/CO/155 2K21/CO/167
Table of Content

Introduction Methodology

Problem Statement Data Flow

Motivation Challenges

Objectives Solutions

Core Technologies Testing Snippets

System Architecture Conclusion


Introduction

The advancements in artificial intelligence (AI) have significantly impacted various industries,
with healthcare emerging as a domain ripe for transformation. The 'Medical Chatbot using
Generative AI' project incorporates the integration of AI technologies with practical applications,
aiming to bridge the gap in accessibility and accuracy of medical information.
Problem Statement

Accessing reliable and immediate medical information is


a persistent global challenge, exacerbated by barriers
such as limited healthcare infrastructure, financial
constraints, and geographical remoteness. Millions of
individuals, especially in underserved regions, face delays
in diagnosis and treatment due to a lack of immediate
professional medical guidance. Furthermore, the digital
age has led to an overwhelming amount of unverified
and misleading health information online.
Motivation

The motivation for this project stems from a growing need to address significant challenges in
the accessibility and reliability of healthcare information. Many individuals face delays in
receiving medical advice due to constraints such as insufficient healthcare infrastructure,
geographical barriers, and financial limitations.
Objectives

Ensure accuracy through Achieve scalability and


Develop an accessible
integration with reliable security
medical chatbot
databases
Core Technologies
OpenAI GPT Pinecone
conversational purposes vector storage and querying

AWS EC2 Docker


scalable hosting containerization

Flask LangChain
backend development task management
System
Architecture
Methodology
Environment Setup
A Python environment managed through Conda to isolate
dependencies. Flask for backend APIs, Pinecone for vector
storage, OpenAI for embeddings, and Docker for deployment.

Data Preparation
Datasets are curated, preprocessed, and transformed into
embeddings stored in Pinecone.

API Development
Flask serves as the backend server managing
communication between components.

Frontend Integration
Interactive user interface will be built using HTML, CSS and
AJAX.
Data Flow

Embedding is Context-aware
User query is
retrieved from response is generated
vectorized
Pinecone by GPT
Challenges
Dataset Quality
The chatbot's effectiveness is limited by the comprehensiveness and
accuracy of the curated medical dataset, requiring ongoing updates to
remain relevant.

Dependency on Pretrained Models


Relying solely on general-purpose GPT models can lead to inadequate
handling of complex or domain-specific medical queries.

Deployment Costs and OpenAI Rate Limits


High operational expenses and API usage restrictions can hinder
scalability and consistent availability of the system.
Solutions
Dataset Quality
Ensure continuous curation and updating of the medical dataset by
integrating trusted and authoritative sources such as medical research
papers, WHO guidelines, or verified FAQs.

Dependency on Pretrained Models


Dependency on pretrained models may lead to overfitting, which may
lead to inaccurate results. Training the model on various datasets
including various domains of medicine such as Pathology, Pharmacology,
Anatomy, etc, will broaden the model training.

Deployment Costs and OpenAI Rate Limits


Purchasing premium tiers of OpenAI and AWS for seamless querying and
hosting.
Output Snippets while Testing

Embedding fetched from Pinecone database


Output Snippets while Testing

Context-aware responses generated by OpenAI


Conclusion

The project will demonstrate the potential of generative AI in transforming


healthcare accessibility and efficiency. It will address critical challenges like
misinformation and accessibility through a scalable, secure, and responsive
chatbot system.
Thank You

You might also like