A Modular IoT-Cloud-AI Architecture for Accessible Cardiac Health Monitoring: Design, Implementation, and Experimental Validation
A modular three-tier IoT-Cloud-AI architecture for accessible cardiac health monitoring is proposed. Three principal contributions are made: (i) a modular architecture enabling independent evolution of hardware, backend, and AI components; (ii) a hybrid risk assessment method combining deterministic rule-based scoring with large language model (LLM)-generated insights; and (iii) a validated open specification supporting reproducibility. A prototype system, Kardia, was implemented using low-cost components (ESP32, MAX30102) and validated experimentally. Heart rate accuracy was within 3 BPM of reference devices, end-to-end latency remained under 2 seconds, and risk stratification was consistent across all test scenarios. The results demonstrate the feasibility of AI-enhanced, home-based cardiac wellness monitoring.