Off-line Heart Rate Error Measurement of Underwater Samples

Underwater sampling of heart rate for sports training has growing attention recently because of the availability of new sensors able to gather data while the user is swimming. Namely, optical sensor for the wrist and strap sensor for the chest. Underwater data transmission is not an option, forcing the analysis to be done off-line. Thus, movement and distance from heart could infuence the gap between data from sensors. This work describe the procedure to calculate: Pearson’s product-moment correlation coeff cient, Lin’s concordance correlation coeff cient, Root-Mean-Square Error, and Dynamic Time Wrapping’s normalized distance over both signals. Mean values for: Pearson’s coeff cient is 0.71, Lin’s coeff cient is 0.66 with bias of 0.92, RMSE is 11.6 bmp, and normalized distance is 1.69. However, results from measurements are scattered, making it diff cult to trust results from the wrist sensor.