Multi-Agent Framework for Resilient Medical Triage in the Internet of Medical Things (IoMT)

Abstract

Remote monitoring has evolved from reactive systems to complex Internet of Medical Things (IoMT) ecosystems. However, centralized architectures face critical vulnerabilities: high latency, absolute dependence on connectivity and its state, and a high rate of false positives in detections due to the system’s own conditions, such as fall detection. This paper presents a distributed multi-agent framework designed to be executed at the edge using ESP32/Single Board Computer (SBC) hardware and MicroPython. The architecture replaces the monolithic logic of a system with a society of autonomous agents (ECG, Falls, Environmental, etc.) that collaborate through the Micro Agent Communication Protocol (μACP). The novelty lies in cooperative reasoning; the system emulates clinical judgment by correlating biological and environmental dimensions through Bayesian fusion. The results project the validation of an accuracy close to 100% in the discrimination of activities of daily living (ADL) and a latency reduction compared to cloud-based models.