Medical RAG ChatBot

An AI-powered medical assistant chatbot that uses Retrieval-Augmented Generation (RAG) to provide accurate medical information based on a comprehensive database of medical cases and doctor responses.

Features

• RAG-based responses: Retrieves relevant medical cases before generating answers
• Vector search: Uses Qdrant for efficient similarity search
• OpenAI integration: Powered by GPT models for intelligent responses
• Document chunking: Efficiently processes large medical datasets
• Token tracking: Monitors API usage and costs
• Interactive chat: Command-line interface for easy interaction

Architecture

┌─────────────────┐
│   User Query    │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Embedding      │
│  (OpenAI)       │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Vector Search  │
│  (Qdrant)       │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  Similar Docs   │
│  Retrieval      │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  LLM Context    │
│  Formation      │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  GPT Response   │
│  Generation     │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  User Response  │
└─────────────────┘

Installation

Prerequisites

• Python 3.12+
• OpenAI API key
• Qdrant account (cloud or local instance)

Setup

1. Clone the repository

   git clone https://github.com/khalilCodeX/MedicalRagChatBot.git
   cd MedicalRagChatBot

2. Download the dataset

   Download the medical chatbot dataset and save it as ai-medical-chatbot.csv in the project root.

   Dataset sources:

   • Kaggle Medical Datasets
   • Or your own medical Q&A dataset in CSV format

   Required CSV columns:

   • Description: Medical case description
   • Patient: Patient query/question
   • Doctor: Doctor's response

3. Set up Python environment

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

4. Install dependencies

   pip install -r requirements.txt

5. Configure environment variables

   Create a .env file in the project root:

   OPENAI_API_KEY=your_openai_api_key_here
   QDRANT_URL=your_qdrant_url_here
   QDRANT_API_KEY=your_qdrant_api_key_here
   QDRANT_COLLECTION_NAME=medical_chatbot

Usage

First-time setup (Embed documents)

Before using the chatbot for the first time, you need to embed the medical documents into the vector database:

# In chat.py, uncomment these lines for first run:
doc_chunks = bot.tokenize_chunkify_documents()
bot.embed_documents(doc_chunks)

Run:

python chat.py

This will:

• Load the CSV dataset
• Chunk documents into manageable pieces
• Create embeddings using OpenAI
• Store vectors in Qdrant

Regular usage

After initial setup, comment out the embedding lines and use:

python chat.py

Then interact with the chatbot:

Enter your medical query (or type 'q' to quit): What are the symptoms of diabetes?
AI Response: [AI-generated response based on retrieved medical cases]

Enter your medical query (or type 'q' to quit): q
Exiting the chat. Goodbye!

Project Structure

MedicalRagChatBot/
├── chat.py              # Main chat interface
├── dataloader.py        # Data loading and document processing
├── genai_llm.py         # LLM chain and prompt management
├── vectordb.py          # Qdrant vector database operations
├── management.py        # OpenAI and Qdrant client setup
├── token_calc.py        # Token counting and cost estimation
├── .env                 # Environment variables (not in git)
├── .gitignore          # Git ignore rules
└── README.md           # This file

Components

dataloader.py

• Loads medical dataset from CSV
• Creates LangChain documents with metadata
• Chunks documents for efficient processing

vectordb.py

• Manages Qdrant vector store
• Embeds documents using OpenAI embeddings
• Performs similarity search for relevant documents

genai_llm.py

• Creates LLM chains with context
• Formats retrieved documents with metadata
• Generates system and user prompts

management.py

• Handles OpenAI client initialization
• Manages Qdrant client connection

token_calc.py

• Calculates token counts for text
• Estimates API costs

Configuration

Document Chunking

Default settings in dataloader.py:

chunk_size=1000  # Characters per chunk
overlap=200      # Overlap between chunks

Vector Search

Default retrieval in vectordb.py:

k=5  # Number of similar documents to retrieve

Embeddings

Default model in vectordb.py:

model="text-embedding-3-small"

Cost Estimation

The chatbot tracks token usage and estimates costs:

• Input tokens: $0.05 per 1M tokens
• Output tokens: $0.40 per 1M tokens (estimated as 2x input)

Disclaimer

IMPORTANT: This chatbot is for educational and informational purposes only. It should NOT be used as a substitute for professional medical advice, diagnosis, or treatment. Always consult qualified healthcare professionals for medical concerns.

License

MIT License - feel free to use and modify for your projects.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Issues

If you encounter any problems, please open an issue on GitHub.

---

Built with LangChain, OpenAI, and Qdrant