MedGuardAI — Real-Time Medical IoT, AI Anomaly Detection & Telemedicine Platform

Licence Graduation Capstone (PFE — USTHB) · Hardware ECG & Blood Glucose Sensors + Raspberry Pi + Python/Flask + Flutter + WebRTC

1. System Overview

MedGuardAI is an end-to-end connected healthcare ecosystem engineered to bridge physical biomedical sensors, real-time deep learning anomaly detection, and clinical telemedicine workflows.

Unlike isolated notebook models, MedGuardAI streams live physiological signals from hardware ECG and blood glucose sensors through a Raspberry Pi edge relay to a Python/Flask inference backend, triggering instant emergency alerts and unlocking WebRTC video consultations between patients and physicians.

---

2. End-to-End Architecture

[Physical ECG & Glucose Sensors]
        │ (Serial / BLE Acquisition)
        ▼
[Raspberry Pi Edge Relay] ──► Real-Time Telemetry Stream
        │
        ▼
[Python / Flask Backend + MongoDB]
  ├─► Deep Learning ECG & Glycemia Anomaly Detection Models (.h5 / Keras)
  ├─► Multi-Tier Clinical Alert Engine (Critical / Warning / Normal)
  ├─► Appointment Scheduling & Automated Digital Prescription Generator
  └─► WebRTC Signaling & Real-Time Chat Server
        │
        ▼
[Cross-Platform Flutter Mobile & Web Dashboards]
  ├─► Patient App: Live ECG waveform, glucose trendlines, emergency SOS, 1-tap video call
  └─► Doctor Portal: Multi-patient telemetry monitor, diagnostic history & prescription builder

---

3. Related Model Repositories

• Deep Learning ECG Anomaly Detection Model
• Deep Learning Blood Glucose Anomaly Detection Model
• Patient Fall Detection Computer Vision Model