---
title: PubMed RAG Chatbot
emoji: 🔬
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.31.0
app_file: app.py
pinned: false
license: mit
short_description: Chatbot that answers questions using PubMed research
tags:
- medical
- rag
- pubmed
- chatbot
- healthcare
- research
- langchain
- openai
---

# 🔬 PubMed RAG Chatbot

An intelligent chatbot that provides research-backed answers to medical questions by searching through a curated database of PubMed abstracts. Built with advanced Retrieval-Augmented Generation (RAG) technology.

## 🎯 What This Does

This application combines the power of medical literature with AI to provide informative responses about:
- Drug mechanisms and side effects
- Disease symptoms and treatments  
- Latest medical research findings
- Clinical procedures and guidelines
- Health conditions and management

## ✨ Key Features

- **🔍 Research-Grounded**: All responses are based on actual PubMed abstracts
- **💬 Conversational**: Maintains context for follow-up questions
- **⚡ Real-time**: Instant responses with source attribution
- **🎨 User-Friendly**: Clean, intuitive chat interface

## 🚀 How to Use

1. **Ask a Medical Question**: Type any health or medical research question
2. **Get Research-Based Answers**: Receive responses grounded in medical literature
3. **Follow Up**: Ask clarifying questions - the bot remembers your conversation
4. **Explore**: Try the example questions to see what's possible

### 💡 Example Questions

- *"What are the side effects of metformin?"*
- *"How does aspirin prevent heart attacks?"*
- *"What are the latest COVID-19 treatment options?"*
- *"Tell me about Type 2 diabetes management"*
- *"What does research say about Mediterranean diet benefits?"*

## 🔧 Technical Details

**Architecture**: Retrieval-Augmented Generation (RAG)
- **Vector Database**: FAISS with PubMed abstracts
- **Embeddings**: HuggingFace `intfloat/e5-base`
- **Language Model**: OpenAI GPT-3.5-turbo
- **Framework**: LangChain + Gradio
- **Memory**: Conversation buffer for context


## 🎨 Interface Features

- **Dark/Light Mode**: Automatic theme adaptation
- **Responsive Design**: Works on desktop and mobile
- **Example Prompts**: Quick-start suggestions
- **Clear Chat**: Reset conversation anytime
- **Loading Indicators**: Visual feedback during processing

## ⚠️ Important Disclaimers

- **Not Medical Advice**: This tool provides information, not medical advice
- **Educational Purpose**: Designed for learning and research
- **Consult Professionals**: Always seek qualified medical advice for health decisions
- **Verify Information**: Cross-check important information with healthcare providers


The app supports various configuration changes:
- Switch embedding models
- Adjust retrieval parameters
- Modify response temperature
- Change UI themes and styling

## 📊 Data Sources

This chatbot searches through a curated database of PubMed abstracts covering:
- Clinical research papers
- Medical case studies
- Drug trials and reports
- Disease research findings
- Treatment effectiveness studies

## 🔄 Updates & Maintenance

- **Model Updates**: Regularly updated with latest research
- **Performance**: Continuously optimized for speed and accuracy
- **Safety**: Regular security and content safety reviews
- **Features**: New capabilities added based on user feedback

## 🤝 Feedback & Support

Found an issue or have suggestions?
- Use the Community tab to report bugs
- Share feedback on response quality
- Suggest new features or improvements
- Report any inappropriate content

## 📋 Dependencies

```txt
gradio>=4.0.0
langchain>=0.1.0
langchain-openai>=0.1.0
faiss-cpu>=1.7.0
sentence-transformers>=2.2.0
openai>=1.0.0
```

## 🏷️ Use Cases

**Researchers**: Quick literature reviews and fact-checking
**Students**: Learning about medical concepts and mechanisms  
**Educators**: Teaching aid with research-backed information
**Healthcare Workers**: Reference tool for clinical information
**General Public**: Health education and awareness

## 🔐 Privacy & Security

- No personal health data is stored
- Conversations are not logged or retained
- OpenAI API calls follow their privacy policy
- No user identification or tracking

---