DeliriumWatch

A real-time environmental and patient monitoring system designed for Grand River Hospital staff, validated through simulated testing.

HEALTHCARE / SIMULATED VALIDATION

Face and room readings

Illustrative demo

Eyes: Closed
Blinks in window: 0
Expression: Relaxed
Room light: 80 lx
Room noise: 34 dB

No example thresholds crossed

The example readings are within the demo’s configured limits.

A synthetic face with example landmarks, sensor readings and thresholds. No camera is used. Flags prompt a staff check-in; they are not a diagnosis or a delirium probability.

Less manual monitoring: 90%+
Faster test response: 40%
Dropped test readings: 0

About this project

Built a real-time Python monitoring pipeline with SQL-backed secure login and role-based access control for hospital staff. During simulated testing, the system reduced manual environmental monitoring by over 90%.

Implemented continuous OpenCV eye detection and a custom TensorFlow and scikit-learn model for automated sleep and blink classification. A live Flask dashboard surfaced abnormal conditions through automatic alert flagging, reducing response time by 40% during validation.

Engineered serial data ingestion and threshold validation for stable multi-sensor tracking with zero dropped readings during testing.

How it works

SENSOR + VIDEOSerial ingestion
CV CLASSIFICATIONEye + sleep detection
STAFF DASHBOARDFlagged conditions

Built with

Raspberry Pi
Arduino
Python
TensorFlow
scikit-learn
OpenCV
Flask
C/C++
HTML
CSS