A comprehensive healthcare application that matches cancer patients to relevant clinical trials using advanced natural language processing and intelligent scoring algorithms. Designed for healthcare professionals, researchers, and medical institutions.
TrialMatchAI combines advanced medical natural language processing with intelligent scoring algorithms to match cancer patients with relevant clinical trials. The application processes natural language patient descriptions, extracts medical entities using biomedical language models, and matches them against a comprehensive database of cancer clinical trials with confidence scoring.
- Medical NLP: Uses PubMedBERT and rule-based extraction for medical entity recognition
- Intelligent Matching: Weighted scoring system prioritizing conditions, demographics, and trial phases
- Confidence Scoring: Provides match confidence percentages and explanations
- Analytics Dashboard: Real-time charts and visualizations of matching results
- Professional Interface: Modern, responsive design for healthcare workflows
- Named Entity Recognition: Automatically identifies medical conditions, demographics, treatments, and lab values
- Categorized Entities: Groups entities into logical categories for better matching
- Fallback System: Robust rule-based extraction when advanced models aren't available
- Entity Visualization: Interactive display of extracted medical entities
- Weighted Scoring: Prioritizes critical fields (conditions, age, sex) over secondary criteria
- Confidence Metrics: Provides match confidence percentages (0-100%)
- Match Explanations: Human-readable explanations for why trials matched
- Fuzzy Matching: Handles variations in medical terminology
- Confidence Distribution: Histogram showing match quality distribution
- Condition Analysis: Bar charts of most common conditions in matches
- Real-time Metrics: Live statistics on total matches, average confidence, and recruiting trials
- Export Functionality: Download results as CSV for further analysis
- Sample Cases: Pre-loaded patient scenarios for quick testing
- Interactive Interface: Modern, responsive design with custom styling
- Error Handling: Comprehensive error handling with informative messages
- Help System: Built-in guidance for optimal usage
- Python 3.8+ - Core programming language
- Streamlit - Modern web application framework
- Transformers - Hugging Face library for medical NLP models
- Plotly - Interactive data visualization
- Pandas - Advanced data manipulation and analysis
- scikit-learn - Machine learning utilities
- PubMedBERT - Biomedical language model for medical entity extraction
Streamlit Cloud Deployment Coming Soon
# Clone the repository
git clone https://github.com/padg9912/TrialMatchAI.git
cd TrialMatchAI
# Install dependencies
pip install -r requirements.txt
# Run the application
streamlit run app.pyThe app will be available at http://localhost:8501
- Enter Patient Data: Use the text area to describe patient demographics and medical conditions
- Choose Sample Cases: Select from pre-loaded patient scenarios or enter custom data
- Analyze Entities: Click "Analyze Entities" to see what medical terms were extracted
- Find Matches: Click "Find Matching Trials" to get ranked results with confidence scores
- Review Results: Explore matching trials with detailed explanations and confidence metrics
The application includes pre-loaded sample cases:
- Breast Cancer Patient:
female, 45 years old, breast cancer, HER2 positive, no prior chemotherapy, non-smoker - Prostate Cancer Patient:
male, 68, prostate cancer, stage II, hypertension, prior surgery - Pediatric Leukemia:
male, 10, acute lymphoblastic leukemia, no CNS involvement, first relapse - Lung Cancer Patient:
female, 62, lung cancer, smoker, stage III, prior radiation therapy - Advanced Melanoma:
male, 55, metastatic melanoma, BRAF positive, immunotherapy naive
- Include age, gender, and specific cancer type
- Mention cancer stage and biomarker status (HER2, EGFR, etc.)
- Include treatment history and smoking status
- Be specific about geographic preferences if relevant
- Medical NER: Automatically identifies medical conditions, demographics, treatments, and lab values
- Categorization: Groups entities into logical categories for better matching
- Visualization: Shows extracted entities in an organized, color-coded display
- Weighted Scoring: Conditions (3x), Demographics (2x), Trial Phases (1.5x), Other fields (1x or less)
- Confidence Scoring: Provides match percentages and explanations
- Fuzzy Matching: Handles medical terminology variations
- Confidence Distribution: Visual histogram of match quality
- Condition Analysis: Bar chart of most common conditions in matches
- Real-time Metrics: Live statistics and KPIs
- Export Options: Download results as CSV for further analysis
TrialMatchAI/
├── app.py # Main Streamlit application
├── models/
│ └── nlp_model.py # Medical entity extraction (PubMedBERT + rules)
├── utils/
│ └── matcher.py # Intelligent trial matching algorithm
├── datasets/
│ └── cancer_studies.csv # Clinical trials database
├── requirements.txt # Python dependencies
└── README.md # This file
- MedicalEntityExtractor: Handles NLP with fallback to rule-based extraction
- TrialMatcher: Implements weighted scoring and confidence calculation
- Streamlit UI: Modern, responsive interface with analytics
- Data Pipeline: Efficient CSV loading and caching
- High Confidence Matches: >70% confidence indicates strong alignment
- Medium Confidence: 40-70% indicates good potential matches
- Low Confidence: <40% indicates partial or weak matches
- Entity Extraction: ~2-3 seconds for complex medical descriptions
- Trial Matching: ~1-2 seconds for 1000+ trial database
- UI Responsiveness: Real-time updates and smooth interactions
- Advanced medical NLP with PubMedBERT integration
- Intelligent weighted scoring system
- Confidence metrics and explanations
- Interactive analytics and visualizations
- Professional UI with sample cases
- Comprehensive error handling
- Streamlit Cloud deployment
- Real-time clinical trial data updates
- Advanced eligibility criteria parsing
- Integration with ClinicalTrials.gov API
- EHR integration via FHIR API
- Multi-language support
- Mobile-optimized interface
- User accounts and saved searches
- Advanced reporting and analytics
- Integration with hospital systems
- Fork the repository
- Create a feature branch (
git checkout -b feature/your-feature) - Commit your changes (
git commit -am 'Add new feature') - Push to the branch (
git push origin feature/your-feature) - Create a Pull Request
MIT License
- ClinicalTrials.gov for open clinical trial data
- Streamlit and Hugging Face for open-source tools
- Biomedical NLP research community for language models
Built for healthcare professionals, researchers, and medical institutions. Contributions and feedback welcome.