All projects
Project 02 · Live · Research · Full-stack
ASAF — AI-Enhanced Smart Antenatal Framework
A full-stack research prototype for cross-device pregnancy risk monitoring by Lady Health Workers operating in low-connectivity rural environments. The system features a trained Random Forest model for risk prediction, a last-write-wins offline sync algorithm, and a strict role-based access control model — all documented in an academic research paper.
FastAPIPython SQLiteRandom Forest ML JWT AuthReact (Babel)
System flow
React SPASingle HTML file
→
FastAPIREST + JWT
→
SQLitePer-user isolation
+
Random Forest86.2% accuracy
+
Offline queueSync on reconnect
AI pregnancy risk prediction
A trained Random Forest model achieves 86.2% accuracy classifying antenatal risk levels from patient vitals. When connectivity is unavailable, the system falls back to an on-device decision tree so health workers can still make risk assessments in the field.
Offline-first sync
Patient registrations and risk examinations are queued locally when offline. On reconnect, POST /api/sync applies them using last-write-wins conflict resolution (Algorithm 1 from the research paper). A simulated field conflict demo is built into the Sync tab.
Production-grade security
Passwords hashed with PBKDF2-HMAC-SHA256 at 100,000 iterations. JWT-based session management with short-lived tokens. Per-user data isolation enforced at the query layer. Full audit log of logins, edits, and syncs across all user types.
Three-tier role system
Admin oversees the entire platform and manages all accounts. Health Workers manage only their own patient records. Patient accounts get a companion view with pregnancy timelines, health tips, and appointment reminders.