Revolutionizing Medical Data Analysis with AI
MedGem is an innovative medical data analysis platform that combines the power of Google's Gemini AI with intuitive data visualization and analysis tools. Built during a hackathon, this project aims to make medical data analysis more accessible, efficient, and insightful for healthcare professionals and researchers.
Demo: Watch on YouTube
- Seamless integration with Google's Gemini AI model
- Natural language processing for medical data queries
- Intelligent data interpretation and insights generation
- Context-aware responses based on medical domain knowledge
- Automated hypothesis generation with supporting evidence
- Three-phase analysis process:
- Data Understanding and Cleaning
- Exploratory Data Analysis and Correlation Analysis
- Hypothesis Formulation
- Support for various medical data formats (CSV, Excel, etc.)
- Secure file upload and storage
- Real-time data processing and analysis
- Interactive data visualization
- Comprehensive data cleaning and preprocessing:
- Missing value handling with multiple strategies (mean, median, mode imputation)
- Data format standardization (dates, times, units)
- Outlier detection and handling
- Data type validation and conversion
- Advanced metadata extraction:
- Basic file information (rows, columns, data types)
- Descriptive statistics (min, max, mean, median, std dev)
- Missing data analysis
- Distribution characteristics
- Relationship exploration
- Modern, responsive web interface built with Next.js
- Intuitive chat interface for data queries
- Real-time feedback and suggestions
- Session management and project organization
- Interactive hypothesis visualization
- Real-time code execution and results display
- Support for iterative analysis and refinement
- Framework: Next.js 15.2.3
- UI: React 19, TailwindCSS
- Data Visualization: Custom components with MathJax support
- Authentication: Supabase Auth
- File Processing: PapaParse, XLSX
- Markdown Support: React Markdown for hypothesis display
- Framework: FastAPI
- AI Integration: Google Gemini API
- Database: Supabase
- Deployment: Docker, Google Cloud Run
- Data Processing: Pandas, NumPy
- Code Execution: Jupyter Notebook environment
- Node.js 18+
- Python 3.8+
- Docker (optional)
- Google Cloud Platform account (for deployment)
- Google Gemini API key
-
Clone the repository:
git clone https://github.com/yourusername/medgem.git cd medgem -
Set up the frontend:
cd frontend npm install npm run dev -
Set up the backend:
cd backend python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install -r docker/requirements.txt export PYTHONPATH=. python app/main.py
-
Set up your environment variables:
# Frontend (.env) NEXT_PUBLIC_SUPABASE_URL=your_supabase_url NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_key # Backend export GOOGLE_API_KEY=your_gemini_api_key
medgem/
├── frontend/ # Next.js frontend application
│ ├── src/ # Source code
│ ├── public/ # Static assets
│ └── package.json # Frontend dependencies
├── backend/ # FastAPI backend service
│ ├── app/ # Application code
│ │ ├── api/ # API endpoints
│ │ ├── core/ # Core business logic
│ │ ├── models/ # Data models
│ │ └── database/ # Database interactions
│ ├── docker/ # Docker configuration
│ └── docs/ # Documentation
└── chatbot_wrapper.py # Gemini AI integration
-
Data Upload and Initial Processing
- Upload medical data files (CSV/Excel)
- Automatic metadata extraction
- Data quality assessment
-
Data Cleaning and Preprocessing
- Missing value handling
- Format standardization
- Outlier detection
- Data type validation
-
Exploratory Data Analysis
- Statistical analysis
- Correlation studies
- Distribution analysis
- Relationship exploration
-
Hypothesis Generation
- AI-powered hypothesis formulation
- Supporting evidence collection
- Analytical method suggestions
- Expected outcomes prediction
We welcome contributions! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.
This project is licensed under the MIT License - see the LICENSE file for details.
- Google Gemini AI team for their powerful API
- The Next.js and React communities for their excellent frameworks
- All contributors and supporters of this project
This project was developed during a hackathon, demonstrating the power of rapid prototyping and modern AI integration in healthcare. It showcases how quickly we can build powerful tools that could potentially transform medical data analysis.