JMIR Publications

ABSTRACT

Background:

The rise of wearable devices has revolutionized personal health monitoring, with heart rate (HR) tracking being one of the most valuable metrics for performance and health assessment. While wearable devices are valuable for measuring HR in real-life settings, continuous validation of their accuracy and reliability accross different activities is essential, as performance may vary due to factors such as wearing position, motion or device updates.

Objective:

This study aimed to validate the accuracy and reliability of the HR measurement of two wearable devices, the Polar Verity Sense (VS) and the Polar Vantage V2 (VV2), using optical photoplethysmography (PPG) technology at various wearing positions on the body (i.e., upper arm, forearm, and wrist) while the study participants engaged in different physical activities and intensities.

Methods:

Sixteen healthy participants (27.4 ± 5.8 years) performed nine activities of varying intensities (lying down to high-intensity interval training). The HR measurements from the VS and VV2 were compared to the gold-standard Polar H10 electrocardiogram (ECG) chest strap. The data were processed to eliminate artifacts and outliers, and validity and reliability were assessed using multiple statistical methods, including systematic bias (mean of differences), mean absolute error (MAE) and mean absolute percentage error (MAPE), Pearson’s product-moment correlation coefficient, root mean squared error (RMSE), ± 5% accuracy, ordinary least squares (OLS) linear regression, concordance correlation coefficient (CCC), within-subject coefficient of variation (WCV), standard error of measurement (SEM), and standardized typical error of the estimate (sTEE).

Results:

The results demonstrated that the arm-worn VS exhibited excellent accuracy, particularly on the upper arm (systematic bias = -0.05 beats per minute (bpm), MAE = 1.43, MAPE = 1.35%), as well as reliability (CCC > 0.98) across most of the activities. The wrist-worn VV2 showed low to very good criterion-related validity (non-dominant wrist: systematic bias = 2.56 bpm; MAE = 6.41 bpm; MAPE = 6.82%), with reliability (CCC) values ranging from 0.24 during picking up objects to 0.97 during post-exercise sitting, which were highly dependent on the level of motion artifacts induced by the different activities. The findings highlight the importance of sensor placement and show that motion significantly affects the accuracy of the HR measurements of wearable devices.

Conclusions:

The VS is recommended as a valid and reliable alternative to ECG-based chest straps. Although the VV2 had lower accuracy and reliability than the VS in this study, it still outperformed other wrist-worn devices from previous studies.