Agentic AIoT: Autonomous Multi-Agent Frameworks for Real-Time Anomaly Detection and Intervention in Remote Cardiac Care

Abstract Remote cardiac care increasingly relies on continuous sensing and timely intervention, yet conventional centralized analytics struggle with latency, scalability, and contextual awareness. This paper presents Agentic AIoT, an autonomous multi-agent framework for real-time anomaly detection and intervention in remote cardiac care. The proposed system deploys cooperative agents across edge, gateway, and cloud layers to perform distributed sensing, adaptive analytics, and coordinated response. Edge agents execute lightweight models for on-device signal quality assessment and preliminary anomaly screening, while coordination agents manage task allocation, confidence aggregation, and escalation. Cloud agents perform deeper temporal modeling and population-level learning, enabling continuous policy refinement. Evaluations on benchmark ECG and multi-modal cardiac datasets, coupled with AIoT simulations, demonstrate robust performance: anomaly detection accuracy of 98.1%, sensitivity of 97.4%, specificity of 98.6%, and an AUC of 0.986. Compared to cloud-only baselines, Agentic AIoT reduces end-to-end response latency by 44.3% and network traffic by 39.8%, while improving intervention precision by 5.2%. The results show that autonomous multi-agent orchestration enhances reliability, responsiveness, and scalability, making Agentic AIoT a promising foundation for next-generation, real-time remote cardiac care systems. Keywords Agentic AIoT, Autonomous Multi-Agent Systems, Remote Cardiac Care, Anomaly Detection, Internet of Medical Things (IoMT), Edge–Cloud Intelligence