PARHELIA: Particle Filter-Based Heart Rate Estimation From Photoplethysmographic Signals During Physical Exercise

The photoplethysmographic (PPG) signal is an important source of information for estimating heart rate (HR). However, the PPG signal could be strongly contaminated by the motion artifact of the subjects, making HR estimation a particularly difficult problem. In this paper, we propose PARHELIA, a PARticle filter-based algorithm for HEart rate estimation using photopLethysmographIc signAls. The proposed method employs a particle filter, and utilizes the simultaneously recorded acceleration signals from a wrist-type sensor, to keep track of multiple HR candidates. This achieves quick recovery from incorrect HR estimations under the strong influence of the MA. Experimental results for a dataset of 12 subjects recorded during fast running showed that the average absolute estimation error was 1.17 beats per minute (BPM) whereas that of the best-known conventional method, JOSS, is 1.28 BPM. Furthermore, the estimation time of PARHELIA is 20 times shorter than JOSS.