🔬 DermKG: Dermatology Knowledge Graph Explorer

An interactive Streamlit application for exploring connections between dermatology diseases, drugs, and symptoms using a Neo4j knowledge graph.

🚀 Features

• Interactive Graph Visualization: Explore concepts and their relationships
• Shortest Path Finding: Discover connections between any two medical concepts
• AI-Powered Insights: Get intelligent descriptions of medical relationships
• Chat Assistant: Ask questions about the knowledge graph
• Responsive Design: Works on desktop and mobile devices

🌐 Live Demo

Visit the deployed app on Streamlit Community Cloud

🛠️ Local Development

Prerequisites

• Python 3.8+
• Neo4j database (local or cloud instance)
• Git

Installation

1. Clone the repository:

git clone https://github.com/yourusername/DermKG.git
cd DermKG

2. Install dependencies:

pip install -r requirements.txt

3. Set up your Neo4j credentials in .streamlit/secrets.toml:

[neo4j]
uri = "neo4j+s://your-neo4j-instance.databases.neo4j.io"
user = "neo4j"
password = "your-neo4j-password"

4. Run the application:

streamlit run app.py

☁️ Deployment on Streamlit Community Cloud

Step 1: Prepare Your Repository

1. Fork this repository to your GitHub account
2. Ensure all files are committed and pushed to your repository

Step 2: Create a Streamlit Community Cloud Account

1. Go to share.streamlit.io
2. Sign up with your GitHub account
3. Connect your GitHub repository

Step 3: Configure Secrets

1. In your Streamlit Community Cloud dashboard, go to your app settings
2. Navigate to the "Secrets" section
3. Add the following secrets:

[neo4j]
uri = "neo4j+s://your-neo4j-instance.databases.neo4j.io"
user = "neo4j"
password = "your-neo4j-password"

Step 4: Deploy

1. Click "Deploy" in your Streamlit Community Cloud dashboard
2. Select your repository and the app.py file
3. Wait for the deployment to complete

Step 5: Test

1. Visit your deployed app URL
2. Verify that the Neo4j connection works
3. Test the graph visualization and chat features

🔧 Configuration

Neo4j Setup

This application requires a Neo4j database with dermatology concepts. The database should have:

• Nodes: Concepts with a name property
• Relationships: Various types connecting the concepts
• Labels: All nodes should have the Concept label

Environment Variables

The app uses Streamlit secrets for configuration:

• neo4j.uri: Neo4j connection URI
• neo4j.user: Neo4j username
• neo4j.password: Neo4j password

Model Configuration

The app uses lightweight language models optimized for cloud deployment:

• Primary: microsoft/DialoGPT-small
• Fallback: gpt2
• Lightest: distilgpt2

If models fail to load, the app falls back to rule-based responses.

📋 Dependencies

• streamlit>=1.28.0 - Web application framework
• neo4j>=5.12.0 - Neo4j database driver
• streamlit-agraph>=0.0.45 - Graph visualization
• transformers>=4.21.0 - AI model loading
• torch>=2.0.0 - Machine learning backend
• accelerate>=0.20.0 - Model optimization
• sentencepiece>=0.1.99 - Tokenization
• protobuf>=3.20.0 - Protocol buffers

🔒 Security

• Database credentials are stored in Streamlit secrets
• No sensitive information is hardcoded
• The app includes error handling for missing credentials

🐛 Troubleshooting

Common Issues

1. "Neo4j credentials not found"

   • Ensure secrets are properly configured in Streamlit Community Cloud
   • Check that the secret keys match exactly: neo4j.uri, neo4j.user, neo4j.password

2. "Failed to connect to Neo4j"

   • Verify your Neo4j instance is running and accessible
   • Check that the URI, username, and password are correct
   • Ensure your Neo4j instance allows remote connections

3. "LLM models not available"

   • This is normal in resource-constrained environments
   • The app will still function with rule-based responses

4. Slow loading

   • First-time model loading may take a few minutes
   • Subsequent loads should be faster due to caching

Performance Tips

• The app uses caching for database connections and model loading
• Graph visualizations are limited to 50 nodes for performance
• Models are optimized for low memory usage

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Neo4j for the graph database
• Streamlit for the web framework
• Hugging Face for the language models
• The dermatology community for domain knowledge

📧 Support

For questions or issues:

• Open an issue on GitHub
• Check the troubleshooting section above
• Review the deployment documentation

---

Note: This application is for educational and research purposes. Always consult medical professionals for health-related decisions.