CareAssist
A clinical dashboard where ESP32 sensor nodes, MQTT, LSTM anomaly detection, and a triage queue all meet in a sub-2s loop.
open projectArchitected a full-stack clinical dashboard with 30+ reusable UI components, processing live vitals from ESP32 nodes via MQTT. Built a real-time NEWS2 scoring engine with Kalman-filtered sensor data and trend-based anomaly detection. Designed a dynamic OPD triage queue using a composite priority algorithm (α·NEWS2 + β·Risk + γ·WaitTime) to eliminate FIFO blind spots — cutting triage assessment time by 40%.
- Next.js
- TypeScript
- FastAPI
- SQLite
- ESP32
- MQTT
- LSTM
- Tailwind