1 of 15

Wireless Ecosystem for Autonomous Agent-based Real-time E-health IOT Device Management.

Huyen-Trang Le (Paris Saclay), Nguyen-Huong-Giang Le (Sorbonne)

CICT Hackathon Competition

Under the supervision of Prof Nazim Agoulmine, Prof. Massinissa Hamidi & Dr. Adnan Imeri (IBISC Lab, Paris Saclay University)

1

2 of 15

2

Proposed general architecture

Main objective (Project’s target): Develop an IEEE 802.15.11 (Wifi standard) Intent-based Wireless BIoAI-powered edge-cloud healthcare IoT system.

CICT Hackathon Competition

3 of 15

3

Cloud-based Controlling Layer

Cloud Layer Target: Develop an intelligent cloud layer for Wireless BioIOT devices efficient deployment & network orchestration.

D4GEN Hackathon Competition

4 of 15

4

Edge-based Detection Layer

Edge Layer Target: Develop an AI model embedded locally on IoT devices.

D4GEN Hackathon Competition

5 of 15

5

Detailed platform

CICT Hackathon Competition

6 of 15

6

Detailed platform

CICT Hackathon Competition

7 of 15

7

Detailed platform

CICT Hackathon Competition

8 of 15

8

Detailed platform

CICT Hackathon Competition

9 of 15

9

Architectural Overview Platform

Modular Architecture for Medical User Intent in IoT Healthcare Ecosystem:

CICT Hackathon Competition

10 of 15

10

Architectural Overview Platform

AI-Assisted System for Medical User Intent in IoT Healthcare Ecosystem:

Medical User Intent Interpretation: Multi-agent CrewAI orchestration framework for translating natural user intents into executable MCP-Based laboratory operations.
Smart IoT Deployment & Management: AI automatically configures and manages medical IoT devices (physical ESP32 devices) for efficient, secure, and energy-aware monitoring.
IOT Plan Validation & Execution: AI agents help IOT system to adapt dynamically to energy consumption and performance in patient fall detection scenario.
Device Real-Time Monitoring & Telemetry: Edge AI by running the AI model (MiniML) locally on ESP32 devices, for detecting health anomalies on patients.

CICT Hackathon Competition

11 of 15

11

Smart Bio-AIOT Prototype

�

CICT Hackathon Competition

12 of 15

12

Smart Bio-AIOT Results

�

CICT Hackathon Competition

Detect Anomalies

ECG Sensor Real Time Update (Single physical device)

LLM in IOT Device Orchestration

(Multiple devices)

13 of 15

13

Cloud-based Sensor Management Algorithm Benchmark

�

CICT Hackathon Competition

14 of 15

14

Edge-based Sensor Anomalies Detection Benchmark

�

CICT Hackathon Competition

Dataset 1

(200000 lines, 6 features)

Dataset 2

(1000 lines, 6 features)

15 of 15

15

Conclusion & Future Direction

�

General Ecosystem: Develop an intelligent hospital monitoring system combining IoT edge devices, AI agents, and cloud-based decision support, demo details in CICTProject Repository.
Sensor Monitoring: Demonstrate the potential of AI-assisted real-time monitoring of hospital environments (patient rooms, corridors) using ESP32-CAM and medical IOT sensors.
Intelligent Context-Aware AI Agent: Integrates AI reasoning and medical knowledge resources to assist healthcare staff in detecting anomalies and making faster decisions.

CICT Hackathon Competition