Medical Chatbot for Doctors - RAG

This is a prototype for a Medical Assistant chatbot, DoctBot, built using the Retrieval-Augmented Generation (RAG) architecture. This chatbot will have all the relevant information about the patient a doctor is consulting and will answer patient-related questions. Doctors can ask general questions, such as a patient summary, or specific queries, like a patient's allergies. The chatbot will generate responses based on the patient's medical history.

Architecture

In this prototype, I have created two containers: frontend and backend, connected via a network. The entire setup is managed using Docker Compose.

Backend:

• A Python Flask application that listens for requests from the frontend container.
• The backend receives a question from the frontend as a string.
• The question is passed to a retriever to fetch relevant documents.
• The retrieved document and the question are sent to the LLM (Large Language Model).
• The LLM generates a response, which is then sent back to the frontend.

Frontend:

• A simple Python script that communicates with the backend container via the command prompt.

Requirements

• Docker
• Git

Steps to Run

git clone https://github.com/adithya1012/RAG_Medical_ChatBot
cd RAG_Medical_ChatBot
docker-compose build --no-cache
docker-compose up

At this point, both containers will be running, but we cannot directly provide input through the command prompt, as Flask occupies the terminal. To interact with the chatbot, we need to enter the frontend container and run the program. Open a new terminal and execute the following commands:

docker exec -it frontend bash
source venv/bin/activate
python3 main.py

Future Work

• Currently, patient data is read from PDFs. This will be replaced with company-specific databases for real-time access to patient records.
• No validation mechanisms are in place at the moment. In the future, RAG-based guardrails (RAGGas) will be added to detect hallucinations in LLM-generated responses.