Realtime Cardiac Health Monitoring Using AI-Powered Embedded Systems

Arrhythmias is characterized as an irregular heartbeat, which leads to high health risk. It leads to fatal outcomes if it is not detected earlier. Manual detection takes a prolonged time and is expensive. This work employed the single-board computer Raspberry Pi 5 for the classification of arrhythmias, which is cost-effective and portable. MIT-BIH datasets were used to train the CNN model, which attained $99 \%$ classification accuracy. The trained model was stored in an $H 5$ file. The work builds a prototype to demonstrate a realworld application of AI in healthcare i.e., heart disease prediction and discusses future enhancement. In a nonhospital environment with normal people, a bio simulator is used to generate abnormal heart rhythms for prediction. The raw ECG signal was extracted from the biosignal simulator. The. h 5 file was loaded into Raspberry Pi permanently, and the raw ECG signal was fed into the Raspberry Pi whenever the prediction needed to be done. The developed system showcased the suitability of real-time arrhythmia detection, providing a reliable and usable outcome for early detection and sustained cardiac monitoring.