🌐 Live Demo: https://environmental-health-risk-map.vercel.app
📊 API: https://your-render-api-link.onrender.com

An interactive geospatial dashboard for visualizing environmental health risk across Southern California ZIP codes, powered by machine learning and CalEnviroScreen 4.0 data.

• Color-coded ZIP code markers by CES 4.0 burden level (green → red)
• Sidebar with health risk indicators, demographics, ethnicity breakdown, and tips for reducing exposure
• ZIP code search
• Real-time health risk predictions (asthma & cardiovascular) via FastAPI
• Support for unseen ZIP codes using geospatial nearest-neighbor estimation
• Health Vulnerability Index (HVI) combining pollution, health outcomes, and socioeconomic factors
• Heatmap visualization for continuous regional risk mapping
• Top 10 most vulnerable ZIP codes ranking

• Node.js 18+
• Python 3.10+
• npm or yarn

npm install
npm run dev

Open http://localhost:5173 in your browser.

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

API will be available at http://127.0.0.1:8000.

npm run build
npm run preview

Datathon2026/
├── backend/
│ ├── Asthma/
│ │ ├── air_quality.py
│ │ ├── asthma_model.py
│ │ ├── asthma_model.ipynb
│ │ ├── asthma_statistics.py
│ │ ├── shap_importance_plot.py
│ │ └── zip_calculations.py
│ ├── Cardiovascular/
│ │ └── cardiovascular_model.py
│ ├── Toxicity/
│ │ └── LBW&Toxicity.ipynb
│ ├── health_index/
│ │ ├── health_index_score.py
│ │ └── text_health.py
│ ├── generate_predicted_zips.py
│ └── main.py
├── data/
│ ├── cardiovascular.csv
│ ├── data.csv
│ ├── Melissa_zipcodes.csv
│ ├── shap_importance.csv
│ └── top_10_percentile_zips.csv
├── frontend/
│ └── src/
│ ├── components/
│ ├── data/
│ ├── utils/
│ ├── App.jsx
│ ├── App.css
│ ├── main.jsx
│ └── index.css
└── models/
├── best_xgb_model.pkl
├── cardiovascular_model.pkl
└── NeuralNetwork.pkl

• XGBoost (Asthma) – Predicts asthma-related health risk using environmental exposure features (PM2.5, ozone, traffic, etc.)
• XGBoost (Cardiovascular) – Estimates cardiovascular disease risk using similar features
• Neural Network – Additional model for toxic release risk estimation, trained alongside the XGBoost models
• Models trained with data cleaning, preprocessing, cross-validation, and hyperparameter tuning

A composite score combining:

• Pollution burden (air quality, toxic releases)
• Health outcomes (asthma, cardiovascular risk)
• Socioeconomic factors (poverty, education, unemployment, housing burden)

Features are normalized and weighted to produce a unified vulnerability score.

1. Geocode unknown ZIP using Melissa dataset
2. Find nearest known ZIP via haversine distance
3. Use nearest ZIP's features as proxy input
4. Run ML inference to estimate risk

• CalEnviroScreen 4.0
• EPA ECHO
• PurpleAir API
• OpenStreetMap
• Melissa ZIP Code Geocoding Database

Environmental health risk is not evenly distributed. Many high-risk communities remain invisible in raw datasets. Healthy Home Audit was built to make these risks accessible, understandable, and actionable for everyone.

• Provides visibility into environmental health risks for vulnerable communities
• Highlights ZIP codes with high pollution and health vulnerability to guide policy and public awareness
• Supports data-driven decisions for local governments, NGOs, and community organizations
• Empowers residents and policymakers with actionable insights and suggestions for improving health outcomes
• Helps identify environmental justice disparities across Southern California

• A city health department identifies top 10 most vulnerable ZIP codes for asthma intervention programs
• Residents explore their local ZIP's health and pollution profile and receive actionable tips to reduce exposure
• Researchers analyze environmental disparities across Southern California