Heartbeat detection based on filter banks and fuzzy inference for u-healthcare

The ubiquitous healthcare (u-healthcare) system refers to mobile healthcare that manages a user's health information; of critical importance in the healthcare system is the accuracy of the measured health data. However, in actual practice, the accuracy changes remarkably according to user motion artifacts in real life. Therefore, this study shows a simplified algorithm for real-time u-healthcare devices to detect the peak points in photoplethysmograph signals despite the existence of motion artifacts. The proposed algorithm consists of filter banks and fuzzy inference. In experiments to evaluate the performance of the proposed algorithm, we used several motion artifacts including finger and wrist movements, arm swinging, and walking; we then compared our results with the performances for the moving average (MA) and scaled Fourier linear combiner (SFLC) methods. The proposed method showed a better performance than the other methods in the experiment when detecting heartbeats.