🩺 MedQueryDoc AI

Intelligent Medical Document Assistant (LLaMA-2)

Chat with your medical PDFs. Ask in plain English. Get answers grounded in your documents.

---

📌 About the Project

This project is an AI-powered medical document question-answering system built using Large Language Models (LLMs) and vector-based information retrieval. It enables users to ask natural language questions and receive accurate, context-aware answers derived from medical PDF documents.

The system works by loading and splitting PDF documents into smaller chunks, generating vector embeddings using a Hugging Face embedding model, and storing them in Pinecone, a scalable vector database. During inference, user queries are matched against relevant document chunks using similarity search, and responses are generated using LLaMA-2 through a retrieval-augmented generation (RAG) pipeline.

This project demonstrates practical implementation of LLMs + LangChain + Vector Databases for real-world AI applications.

---

🧠 Key Features

• PDF-based medical document ingestion
• Vector similarity search using Pinecone
• Retrieval-Augmented Generation (RAG) pipeline
• Local LLaMA-2 model inference
• Interactive web interface using Flask
• Modular and extensible architecture

---

⚡ How It Works

PDF  →  Embeddings  →  Pinecone  →  LLaMA-2  →  Answer

Phase	Command	What happens
1. Index once	python store_index.py	Loads PDFs, splits text into chunks, generates embeddings, stores vectors in Pinecone
2. Ask anytime	python app.py	Retrieves the most relevant chunks and generates an answer with LLaMA-2

---

🛠️ Built With

• Python
• LangChain
• Flask
• Meta LLaMA-2
• Pinecone (Vector Database)
• Hugging Face Embeddings

---

🚀 Getting Started

Follow the steps below to set up and run the project locally.

---

⚙️ Installation Steps

1️⃣ Clone the Repository

git clone https://github.com/<your-username>/MedQueryDoc-AI-Intelligent-Medical-Document-Assistant.git
cd MedQueryDoc-AI-Intelligent-Medical-Document-Assistant

---

2️⃣ Create a Virtual Environment (Recommended)

If Conda is installed:

conda create -p ./venv python=3.10 -y
conda activate ./venv

Or using Python venv:

python -m venv venv
venv\Scripts\activate        # Windows
source venv/bin/activate     # macOS / Linux

---

3️⃣ Install Dependencies

pip install -r requirements.txt

---

🔐 API Key Setup

Create a .env file in the root directory and add your Pinecone credentials:

PINECONE_API_KEY=your_pinecone_api_key_here

⚠️ Note: Pinecone no longer requires an environment variable (PINECONE_API_ENV) in the new SDK.

---

📄 Add Medical Documents

Create a data/ folder in the project root and add your PDF files:

data/
├── medical_doc_1.pdf
├── medical_doc_2.pdf

---

🧠 Model Setup

Download the quantized LLaMA-2 model and place it inside the model/ directory:

Model: llama-2-7b-chat.ggmlv3.q4_0.bin
Source: https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGML

---

🧬 Index Documents into Pinecone

Before running the chatbot, index the documents:

python store_index.py

This step:

• Loads PDFs
• Splits text into chunks
• Generates embeddings
• Stores vectors in Pinecone

---

▶️ Run the Application

python app.py

Then open your browser and visit:

http://localhost:8080

---

🏗️ Project Architecture

├── app.py                 # Flask application
├── store_index.py         # PDF ingestion & vector indexing
├── src/
│   ├── helper.py          # PDF loading, chunking, embeddings
│   ├── prompt.py          # Prompt template
├── model/
│   └── llama-2-7b-chat.ggmlv3.q4_0.bin
├── templates/
│   └── chat.html
├── static/
│   └── style.css
├── assets/
│   └── social-preview.png # Demo banner used in this README
├── data/                  # PDF documents
└── .env

---

🚧 Future Enhancements

• Multilingual medical support
• Source citation in answers
• Document upload via UI
• Streaming responses
• Cloud-hosted inference
• Authentication & access control

---

⚠️ Disclaimer

This project is for informational and educational purposes only. It is not a medical device and is not a substitute for professional medical advice, diagnosis, or treatment. Always consult a qualified healthcare professional.

---

📜 License

This project is licensed under the GNU General Public License v3.0. See the LICENSE file for more details.

---

⭐ Acknowledgements

This project builds upon open-source tools and research in the fields of LLMs, NLP, and vector databases. Special thanks to the open-source community for enabling practical AI innovation.

---

If you found this project useful, please ⭐ star the repo!