Cooperative PPG/ECG Wearable System for Atrial Fibrillation Diagnosis

Recent advancements in wearable technology, particularly wrist-type electrocardiography (ECG) devices, offer a promising alternative for atrial fibrillation (AF). These devices allow for continuous monitoring but typically require the user to touch an electrode to capture accurate readings. This study introduces a novel photoplethysmography (PPG)/ECG cooperative wearable device that leverages PPG for continuous heart rate (HR) monitoring and ECG for precise AF detection. Beyond sensor development, this study emphasizes the processing of sensor data, with the system utilizing a robust motion artifact (MA) cancellation algorithm and an AF-finite-state machine (AF-FSM) framework to enhance the accuracy of PPG signal analysis. The system employs PPG for preliminary AF screening, and, upon detecting irregularities, it prompts ECG measurements with data transmitted to medical professionals for confirmation. In this study, to enhance the accuracy of AF confirmation while reducing false alarms and increasing true positives, we mainly propose a robust MA cancellation algorithm combined with an AF-FSM framework. Evaluation using the BAMI-II and ISPC datasets shows that our method achieves a mean absolute error (MAE) of 1.31 and 1.16 beats per minute (bpm), respectively, for HR estimation and an AF detection sensitivity of 0.9927, specificity of 0.9768, and accuracy of 0.9847 on a clinical dataset. On the MIMIC PERform AF dataset, our method achieved an AF detection sensitivity of 0.9536, specificity of 0.9500, and accuracy of 0.9517. A pilot study further validates the system's practical application in a clinical setting, demonstrating its potential for long-term, user-friendly cardiac monitoring.