📅 AI Medical Assistant Chatbot — RAG-based Application

🎥 Watch the full project walkthrough: YouTube Video

🗂️ Architecture PDF: View Architecture

---

🧠 Project Overview

This application is a Medical Domain Chatbot built using Retrieval-Augmented Generation (RAG). It allows users to upload their own medical documents (e.g., textbooks, reports), and the system intelligently answers queries by retrieving the most relevant content before generating a final response.

---

🎓 What is RAG?

RAG (Retrieval-Augmented Generation) enhances language models by supplying relevant external context from a knowledge base, preventing hallucinations and improving accuracy, especially for factual or specialized domains like medicine.

---

🔄 Architecture

User Input
   ↓
Query Embedding → Pinecone Vector DB ← Embedded Chunks ← Chunking ← PDF Loader
   ↓
Retrieved Docs
   ↓
     RAG Chain (Groq + LangChain)
   ↓
LLM-generated Answer

For a detailed view, refer to the MedicalAssistant.pdf

---

📚 Features

• Upload medical PDFs (notes, books, etc.)
• Auto-extracts text and splits into semantic chunks
• Embeds using Google/BGE embeddings
• Stores vectors in Pinecone DB
• Uses Groq's LLaMA3-70B via LangChain
• FastAPI backend with endpoints for file upload and Q&A

---

🌐 Tech Stack

Component	Tech Used
LLM	Groq API (LLaMA3-70B)
Embeddings	Google Generative AI / BGE
Vector DB	Pinecone
Framework	LangChain
Backend	FastAPI
Deployment	Render

---

📚 API Endpoints

POST /upload_pdfs/ --- Upload one or more PDF files

POST /ask/ --- Ask a question --- Form field: `question`

---

📁 Folder Structure

└── 📁assets
    ├── DIABETES.pdf
    ├── MedicalAssistant.pdf
    └── medicalAssistant.png

└── 📁client
    └── 📁__pycache__
        ├── config.cpython-311.pyc
    └── 📁components
        └── 📁__pycache__
            ├── chatUI.cpython-311.pyc
            ├── history_download.cpython-311.pyc
            ├── upload.cpython-311.pyc
        ├── chatUI.py
        ├── history_download.py
        ├── upload.py
    └── 📁utils
        └── 📁__pycache__
            ├── api.cpython-311.pyc
        ├── api.py
    ├── app.py
    ├── config.py
    └── requirements.txt

└── 📁server
    └── 📁__pycache__
        ├── logger.cpython-311.pyc
        ├── main.cpython-311.pyc
        ├── test.cpython-311.pyc
    └── 📁middlewares
        └── 📁__pycache__
            ├── exception_handlers.cpython-311.pyc
        ├── exception_handlers.py
    └── 📁modules
        └── 📁__pycache__
            ├── llm.cpython-311.pyc
            ├── load_vectorstore.cpython-311.pyc
            ├── query_handlers.cpython-311.pyc
        ├── llm.py
        ├── load_vectorstore.py
        ├── pdf_handlers.py
        ├── query_handlers.py
    └── 📁routes
        └── 📁__pycache__
            ├── ask_question.cpython-311.pyc
            ├── upload_pdfs.cpython-311.pyc
        ├── ask_question.py
        ├── upload_pdfs.py
    └── 📁uploaded_docs
        ├── DIABETES.pdf
        ├── Supratim Nag - LOR.pdf
    ├── .env
    ├── logger.py
    ├── main.py
    ├── requirements.txt
    └── test.py

---

⚡ Quick Setup

# Clone the repo
$ git clone https://github.com/snsupratim/medicalAssistant.git
$ cd medicalAssistant/server

# Create virtual env
$ uv venv
$ .venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
$ uv pip install -r requirements.txt

# Set environment variables (.env)
GOOGLE_API_KEY=...
GROQ_API_KEY=...
PINECONE_API_KEY=...

# Run the server
$ uvicorn main:app --reload --port 8000


$ cd medicalAssistant/client

# Create virtual env
$ uv venv
$ .venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
$ uv pip install -r requirements.txt

# Run the server
$ streamlit run app.py

---

🌐 Deployment

• Hosted on Render

• Configure start command as:

  uvicorn main:app --host 0.0.0.0 --port 10000

---

🌟 Credits

• Built by Supratim Nag
• Inspired by LangChain, Groq, Pinecone, and FastAPI ecosystems

---

🎉 License

This project is licensed under the MIT License.