doniljaison/MaternaGuard

MaternaGuard is a maternal health risk triage platform for low-resource settings. It combines a trained ML risk model, a FastAPI backend, and a React web app to help identify high-risk pregnancy cases early and support faster escalation. # MaternaGuard MaternaGuard is a maternal health risk triage platform for low-resource settings. It combines a trained ML risk model, a FastAPI backend, and a React web app to help identify high-risk pregnancy cases early and support faster escalation. ## Live Deployment 1. Frontend (Vercel): maternaguard.vercel.app 2. Frontend preview domain: maternaguard-hwpqljnhp-doni… 3. Backend (Render): maternaguard-backend.onrend… 4. Backend health check: maternaguard-backend.onrend… How to open the deployed app: 1. Open maternaguard.vercel.app in a browser. 2. Go to Log and submit vitals. 3. The frontend calls the Render backend for prediction. 4. View risk result, dashboard updates, and alerts. ## IMPACT AND BENEFITS **70,000+** Maternal deaths in India/year **67%** Preventable with early detection **< 5s** Time to triage result **800M+** Underserved Indians in rural **SDG-3 Aligned • Good Health & Well-being • Ayushman Bharat Integration Ready • NHM Programme Compatible** ### Social & Health Impact - **Saves lives** — flags high-risk before crisis, not after - **Empowers ASHA workers** with clinical AI on basic Android - **Fills last-mile gap** — no hospital infrastructure needed - **Auto SMS referral** eliminates manual escalation delay ### Economic & Scalability - **₹0 Zero-cost stack** — free-tier tools, no licensing fees - **Scales nationally** via REST API + Ayushman Bharat stack - **Reduces unnecessary referrals** — saves transport & tertiary costs - **Open-source** — publishable for global maternal health research - **Sustainable:** runs on existing phones, no new hardware needed ## What The System Does Input vitals used in app prediction flow: 1. Age 2. Systolic BP 3. Diastolic BP 4. Blood Sugar 5. Body Temperature 6. Heart Rate The system predicts one of three classes: 1. Low Risk 2. Mid Risk 3. High Risk For high-risk predictions, the app records an in-app v …