Real-Time Heart Anomaly Detection and Patient Monitoring Using AIoT

Abstract

Heart diseases remain a major cause of death in the global community, especially in the rural and underserved population where prompt medical relief is scarce. It is important to detect and regularly monitor heart conditions, as early identification can help prevent serious consequences. To support such requirement, this paper introduces a real-time, AIoT-based heart monitoring platform composed of low-cost hardware and intelligent software. The pulse oximeter MAX30100 allows continuous monitoring of vital signs using the ESP8266 NodeMCU application on the Raspberry Pi platform: heart rate (BPM) and blood oxygen saturation (SpO2). Transmission of data is wireless and can be tracked remotely and continuously, with data being transmitted to a cloud platform. XGBoost is trained on the UCI Cleveland dataset based on several handpicked features to make risk prediction on heart disease. Optuna, an optimization framework for hyperparameters, is used to tweak the model to be more accurate. This energy-efficient system is particularly admissible in the field implementation at the rural communities, which should positively influence the preventive cardiac care.