Influence of Skin Tone on Accuracy of Wrist-Worn Photoplethysmographic Heart Rate During Exercise: A Pilot Study

Wearable technology that uses photoplethysmography (PPG) to estimate heart rate (HR) by measuring light reflectance at the epidermis is attractive because it is both inexpensive and noninvasive. However, higher melanin content may reduce the amount of light passing through the epidermis. Previous studies examining the impact of skin color on PPG HR measurement accuracy have produced mixed results, however, none objectively measured epidermal melanin. PURPOSE: To investigate the association between objectively-measured melanin content and accuracy of PPG HR measurement across a range of exercise intensities. METHODS: A handheld colorimeter noninvasively measured skin color in 14 healthy adults (mean ± SD; age = 23 ± 5 y; White, n = 7; Black, n = 6; Middle Eastern, n = 1), and was then used to calculate Individual Typology Angle (ITA°), a strong correlate of epidermal melanin. After recording resting HR, participants then cycled continuously in temperate conditions (22 °C, 40% relative humidity) for 10 min each at 50%, 60%, 70%, and 80% of age-predicted maximal HR, respectively, followed by a 10-min seated recovery. Chest-worn HR (criterion) and wrist-worn PPG HR were recorded continuously and summarized into 30-s epochs for analysis. Bland-Altman analysis evaluated device performance. Repeated measures correlation (rrm) quantified the overall within-individual relationship between criterion and PPG HR, and a linear mixed effects model determined whether ITA° influences MAE across rest, exercise, and recovery. An a priori threshold for acceptable mean absolute error (MAE) between criterion and PPG HR was set at ≤5 bpm. RESULTS: Overall, mean bias was −0.1 [95% limits of agreement: ±4.4] bpm; criterion and PPG HR values were highly correlated (rrm = 0.997, P < 0.001). Exercise intensity (β = −0.2, P = 0.04), not ITA° (β = −0.001, P = 0.84), significantly predicted MAE. CONCLUSIONS: We observed a high level of agreement between criterion and PPG HR in this sample, with acceptable overall agreement between devices. Further, MAE decreased with increasing exercise intensity. A larger and more diverse sample is needed to further elucidate whether epidermal melanin, represented as ITA°, is associated with PPG HR accuracy.