Cancer Treatment Assistant 🏥

A comprehensive web application built with Flask and BioBERT that provides intelligent information about cancer treatments, side effects, and medical guidance using advanced NLP processing and AI-powered question answering.

✨ Features

• AI-Powered Question Answering: BioBERT model trained on biomedical literature for accurate cancer-related responses
• Interactive Chat Interface: Modern web-based chat with real-time question answering
• Advanced NLP Pipeline: Automatic spaCy model downloading, multilingual support (English/French), and intelligent text processing
• Semantic Context Retrieval: Sentence transformer-based similarity matching for relevant context extraction
• Comprehensive Knowledge Base: Curated cancer Q&A dataset with medical information
• Modern Web Interface: Bootstrap-powered responsive design with Chart.js visualizations
• Medical Statistics Dashboard: Analytics and insights about cancer treatments and data
• Robust Error Handling: Comprehensive test suite and error management
• Cross-Platform Compatibility: Universal path handling for Windows, macOS, and Linux

🚀 Quick Start

Prerequisites

• Python 3.8 or higher
• pip package manager
• 4GB+ RAM (for BioBERT model)

Installation

1. Clone the repository:

   git clone https://github.com/SecurDrgorP/CancerCare-AI.git
   cd CancerCare-AI

2. Install dependencies:

   pip install -r requirements.txt

3. Download and save the BioBERT model locally (REQUIRED - run this first):

   python scripts/saveModel.py

   This will download the BioBERT model (~1.3GB) and save it locally in the model/ directory.

4. spaCy models will be automatically downloaded on first run (no manual installation needed)

5. Run the Flask application:

   python app.py

   or

   flask run

6. Open your browser and navigate to http://localhost:5000

📁 Project Structure

CancerCare-AI/
├── app.py                    # Main Flask application with routes
├── biobert_qa.py            # BioBERT question answering implementation
├── nlp_pipeline.py          # NLP processing with automatic spaCy downloading
├── context_provider.py      # Semantic context retrieval using sentence transformers
├── data_handler.py          # Data management and processing
├── requirements.txt         # Python dependencies
├── scripts/                 # Setup and utility scripts
│   └── saveModel.py         # BioBERT model download and setup script
├── model/                   # BioBERT model storage
│   └── biobert_v1.1_pubmed_squad_v2_local/  # Local BioBERT model
├── data/                    # Medical datasets
│   └── cancer_qa_dataset.json    # Cancer Q&A knowledge base
├── templates/               # HTML templates
│   ├── base.html           # Base template with Bootstrap
│   ├── index.html          # Landing page
│   ├── chat.html           # Chat interface
│   ├── statistics.html     # Analytics dashboard
│   └── error pages (404.html, 500.html)
├── static/                  # Static web assets
│   ├── css/                # Custom stylesheets
│   └── js/                 # JavaScript files
├── tests/                   # Comprehensive test suite
│   ├── test_app.py         # Flask app tests
│   ├── test_biobert_qa.py  # BioBERT tests
│   ├── test_nlp_pipeline.py # NLP pipeline tests
│   └── test_context_provider.py # Context retrieval tests
└── README.md               # Project documentation

💡 Usage Examples

Web Interface

1. Landing Page (/): Overview with example queries and statistics
2. Chat Interface (/chat): Full-page chat with AI-powered responses
3. Statistics Dashboard (/statistics): Analytics and treatment insights

API Endpoints

• POST /api/chat: Send questions and receive AI-generated answers
• GET /api/statistics: Retrieve treatment and cancer statistics

Sample Queries

• "What are treatment options for breast cancer stage 2?"
• "Side effects of chemotherapy?"
• "Diet recommendations during radiation?"
• "Recovery time after surgery?"
• "What is immunotherapy for lung cancer?"
• "How does radiation therapy work?"
• "Symptoms of ovarian cancer?"
• "Cost of cancer treatments?"

Key Components

1. Flask Application (app.py)

   • RESTful API with multiple routes
   • Template rendering with Jinja2
   • Integration with AI models and data handlers
   • Error handling and logging

2. BioBERT QA System (biobert_qa.py)

   • Loads pre-trained BioBERT model from local storage
   • Performs extractive question answering
   • Optimized for biomedical text understanding
   • GPU/CPU compatibility with automatic device detection

3. NLP Pipeline (nlp_pipeline.py)

   • Automatic spaCy model downloading (en_core_web_sm, fr_core_news_sm)
   • Multilingual text processing (English/French)
   • Language detection and normalization
   • Medical entity extraction and text cleaning

4. Context Provider (context_provider.py)

   • Sentence transformer-based semantic search
   • Finds most relevant context from knowledge base
   • Uses all-MiniLM-L6-v2 model for embeddings
   • Efficient similarity matching and ranking

5. Data Handler (data_handler.py)

   • Manages medical datasets and knowledge base
   • Provides structured access to cancer information
   • Statistics generation and data aggregation

🧪 Testing

The project includes a comprehensive test suite covering all major components.

Running Tests

1. Run all tests:

   pytest

2. Run tests with verbose output:

   pytest -v

3. Run specific test file:

   pytest tests/test_biobert_qa.py

4. Run tests with coverage report:

   pytest --cov=. --cov-report=html

Test Structure

• test_biobert_qa.py: Tests BioBERT model initialization, question answering, and error handling
• test_nlp_pipeline.py: Tests NLP processing, language detection, and spaCy model management
• test_context_provider.py: Tests semantic search, embedding generation, and context retrieval
• test_data_handler.py: Tests data loading, management, and statistics generation

Test Coverage

The test suite covers:

• ✅ Model loading and initialization
• ✅ Question answering accuracy
• ✅ NLP pipeline processing
• ✅ Context retrieval functionality
• ✅ Data handling and management
• ✅ Error handling and edge cases
• ✅ API endpoint responses

📊 Data Sources

The application uses curated medical datasets including:

• Cancer Types: 10+ common cancer types with staging and symptoms
• Treatments: 10+ treatment modalities with effectiveness and costs
• Side Effects: 20+ common side effects with frequencies
• FAQ: 20+ frequently asked questions with medical answers

🔧 Technical Stack

• Backend: Flask web framework with Jinja2 templating
• AI/ML: BioBERT (Transformers), Sentence Transformers, PyTorch
• NLP: spaCy (with automatic model downloading), NLTK, langdetect
• Frontend: HTML5, Bootstrap 5, Chart.js, Custom CSS/JavaScript
• Data Processing: Pandas, NumPy for data manipulation
• Visualization: Chart.js for interactive web charts, Matplotlib/Seaborn for analytics
• Testing: pytest with comprehensive test coverage
• Dependencies: See requirements.txt for complete list
• Model Storage: Local BioBERT model (~1.3GB) for offline operation

⚠️ Medical Disclaimer

IMPORTANT: This application provides general information for educational purposes only. It is NOT a substitute for professional medical advice, diagnosis, or treatment. Always consult with qualified healthcare professionals for medical decisions.

🎯 Success Criteria

• ✅ AI-powered question answering using BioBERT for biomedical accuracy
• ✅ Modern, responsive web interface with professional medical design
• ✅ Comprehensive test suite with good coverage
• ✅ Automatic dependency management (spaCy models)
• ✅ Cross-platform compatibility (Windows, macOS, Linux)
• ✅ Proper medical disclaimers throughout the application
• ✅ Fast response times with local model inference
• ✅ Semantic context retrieval for relevant information

🚀 Future Enhancements

• Enhanced AI Models: Integration with newer biomedical language models
• Real-time Data: Connection to medical databases and research APIs
• Personalized Responses: User profile-based recommendations
• Multi-language Support: Expansion beyond English and French
• Mobile App: React Native or Flutter mobile application
• Voice Interface: Speech-to-text and text-to-speech capabilities
• Clinical Integration: EHR integration and clinical decision support
• Advanced Analytics: Machine learning for treatment outcome predictions

🔧 Troubleshooting

Common Issues

1. Model Download Issues:

   # If saveModel.py fails, try manually:
   python scripts/saveModel.py

2. spaCy Model Problems:

   # Manual installation if auto-download fails:
   python -m spacy download en_core_web_sm
   python -m spacy download fr_core_news_sm

3. Memory Issues:

   • Ensure at least 4GB RAM available
   • Close other applications when running BioBERT
   • Consider using smaller batch sizes

4. Port Already in Use:

   # Use a different port:
   export FLASK_RUN_PORT=5001
   flask run

5. Import Errors:

   # Reinstall dependencies:
   pip install -r requirements.txt --force-reinstall

Testing Issues

If tests fail, check:

• All dependencies are installed: pip install -r requirements.txt
• BioBERT model is downloaded: python scripts/saveModel.py
• Python version compatibility (3.8+)

📝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

📞 Support

For questions or support, please refer to the documentation or create an issue in the repository.

---

Built with ❤️ for cancer patients and caregivers