Cardiology RAG – Retrieval-Augmented Generation with PDFs

This project demonstrates how to build a Retrieval-Augmented Generation (RAG) pipeline for question answering on a collection of cardiology research articles in PDF format. Full Linkedin article at Building a Retrieval-Augmented Generation (RAG) System for Cardiology Research: A Step-by-Step Guide

The system uses LangChain, ChromaDB, and a local Hugging Face model (Qwen-7B-Instruct by default) to retrieve relevant text chunks and generate grounded answers.

---

✨ Features

• Parse multiple PDF documents (e.g., cardiology articles)
• Split text into chunks for efficient retrieval
• Store embeddings in a persistent ChromaDB vector database
• Run a local LLM (no API key required)
• Answer questions with sources included for transparency

---

📂 Project Structure

cardiology_rag/
│── articles/                # Folder with cardiology PDFs
│    ├── atrial_fibrillation.pdf
│    ├── heart_failure.pdf
│    └── coronary_disease.pdf
│── chroma_db/               # Vector DB (auto-created)
│── main.py                  # Main RAG pipeline script

---

⚙️ Installation

Create a clean environment and install dependencies:

# Clone repository
git clone https://github.com/yourusername/cardiology_rag.git
cd cardiology_rag

# Create virtual environment
python -m venv rag_env
source rag_env/bin/activate   # Linux / Mac
rag_env\Scripts\activate      # Windows

# Install dependencies
pip install -r requirements.txt

If you don’t have a requirements.txt yet, install manually:

pip install langchain langchain-community langchain-huggingface langchain-chroma chromadb pypdf sentence-transformers transformers torch accelerate

---

🚀 Usage

1. Place your cardiology PDFs inside the articles/ folder.
2. Run the main script:

python main.py

Example output:

Loaded 90 pages from 10 PDF files.
Total chunks: 634

=== Answer ===
Atrial fibrillation can cause stroke, heart failure, and increased mortality.

=== Sources ===
[1] atrial_fibrillation.pdf | p.12
[2] heart_failure.pdf       | p.4

---

🔧 Customization

• Change model:
  Default is Qwen/Qwen2.5-7B-Instruct. You can try smaller models like google/flan-t5-base or larger ones like mistralai/Mistral-7B-Instruct-v0.2.

• Adjust retrieval:
  Modify k in db.as_retriever(search_kwargs={"k": 4}) to return more or fewer chunks.

• Expand knowledge base:
  Add more PDFs (guidelines, textbooks, papers) into articles/.

---

📌 Applications

• Medical students preparing for exams
• Clinicians reviewing updated guidelines
• Researchers summarizing across papers
• Startups building intelligent assistants

⚠️ Disclaimer: This project is for educational purposes only. Always verify medical information with trusted clinical sources before applying it in practice.

---

📜 License

MIT License – see LICENSE for details.
Copyright (c) 2025 pvrego