EqualCare

EqualCare is a web application that helps identify gender bias in healthcare datasets before they are used in machine learning models or clinical research. It provides a simple upload interface for researchers and developers to analyze their datasets for gender representation imbalance, and offers AI-powered insights on how such imbalance could impact real-world outcomes.

Features

• Upload CSV datasets securely
• Automatically detect and normalize gender data
• Calculate gender breakdown and bias levels
• Visualize data distributions (bar/pie chart support coming soon)
• Generate plain-English explanations for detected bias using RAG (retrieval-augmented generation)
• Maintain a history of uploaded files
• Supports structured output with issue, impact, and solution sections

Tech Stack

• Frontend: React (JavaScript)
• Backend: FastAPI (Python)
• Data Analysis: Pandas
• AI Inference: OpenRouter API (OpenChat 7B via RAG)
• Embeddings: SentenceTransformer (all-MiniLM-L6-v2)
• Vector DB: ChromaDB
• Storage: Supabase (buckets)
• PDF Loader: LangChain Community (PyPDFLoader)

How It Works

1. Users upload a CSV file containing healthcare data.
2. The backend processes the data using Pandas, normalizing gender values and analyzing imbalance.
3. A bias level is computed based on percentage thresholds.
4. The app performs a vector similarity search on related research papers.
5. It sends a structured prompt to a hosted LLM via OpenRouter to generate:
   • Issue: What the imbalance is
   • Impact: Why it matters in real-world contexts
   • Solution: How to address or mitigate it
6. The frontend displays the breakdown and AI-generated explanation.

Setup Instructions

1. Clone the Repository

git clone https://github.com/your-username/equalcare.git
cd equalcare

2. Backend Setup

• Create and activate a virtual environment
• Install dependencies

pip install -r requirements.txt

• Create a .env file with the following keys:

OPENROUTER_API_KEY=your_openrouter_key
SUPABASE_URL=your_supabase_url
SUPABASE_KEY=your_supabase_service_role_key
SUPABASE_BUCKET=your_supabase_bucket_name

• Start the server:

uvicorn main:app --reload

3. Frontend Setup

cd frontend
npm install
npm run dev

To build for production:

npm run build

Deployment

For deployment as a unified full-stack app:

• Build the React frontend (npm run build)
• Serve the /dist folder with FastAPI using StaticFiles
• Configure environment variables in your deployment environment
• Optional: Use Docker for containerized deployment

Folder Structure

/frontend           # React frontend app
/backend            # FastAPI backend (main app.py lives here)
/research_papers    # PDF docs used for RAG vector store
/chroma_db          # ChromaDB persistent vector DB

Roadmap

• Add charting (bar/pie visualization of gender distribution)
• Add auto-detection of dataset categories (e.g., Stroke, Heart Attack)
• Expand AI reasoning to support multiple bias types
• Support multi-class gender labels
• PDF export or report download

License

This project is open-source and available under the MIT License.