RealTime Heart Rate Monitoring Using Photoplethysmographic (PPG) Signals During Intensive Physical Exercises

Heart Rate (HR) is a fundamental vital sign, monitoring which provides essential information for automated healthcare systems. The emerging technology of Photoplethysmograph (PPG) is shown as a feasible candidate for such applications; however, Motion Artifacts (MA) hinder efficient HR estimation using PPG, especially in situations involving physical activities. It is previously shown that even in the presence of sever MA, HR is still traceable with the help of simultaneous acceleration data although at high computational expenses. In this paper, we propose a novel framework, that not only improves the accuracy in HR estimation, but also achieves realtime performance by significantly reducing the complexity of system; mainly due to alleviation of the need for computationally demanding MA cancellation methods. Utilizing an spectrum estimation model (autoregressive) that suits well to the inherent PPG generation process, and benefiting from further intrinsic properties of the environment (e.g., the venous pulsation phenomenon); our framework achieves realtime and delayed (post-processed) Average Absolute Errors (AAE) of 1.19 and 0.99 Beats Per Minute (BPM) respectively, on the 12 benchmark recordings in which subjects run at speeds of up to 15 km/h maximum. Moreover, the system makes standalone implementation feasible by processing input frames (2 channel PPG and 3D ACC) in < 0.004 times of the frame duration, operating on a 3.2 GHz processor. This study provides wearable healthcare technologies with a robust framework for accurate HR monitoring; at considerably low computational costs.