Back to Projects
Hybrid System

ALARMS

IoTPredictiveHeart-healthMonitoring

Problem Statement

Healthcare monitoring systems often lack real-time alerting and secure data handling for critical patient data.

System Architecture

01

ESP32

02

MQTT

03

Flask REST APIs

04

ML Inference

05

WebSockets

06

Real-time Alerts

Solution Overview

Created an end-to-end vitals anomaly detection pipeline (heart rate, body temperature, accelerometer) achieving AUC 0.98 and 94% validation accuracy.

End-to-end vitals anomaly detection pipeline with AUC 0.98
94% validation accuracy on health metrics
Real-time ingestion and alerting using MQTT + Flask REST APIs
Continuous sensor streaming from ESP32 with low-latency monitoring
Monitoring UI with live updates via WebSockets
94%

accuracy

0.98

auc

<100ms

latency

Technology Stack

PythonFlaskTensorFlow/KerasMQTTESP32WebSocketsAWSNext.js