Influence of skin pigmentation on the accuracy and data quality of photoplethysmographic heart rate measurement during exercise.

Abstract

Introduction: Wearable devices often use photoplethysmography to estimate heart rate (HR) by measuring light reflected from the cutaneous vascular bed. Notably, higher melanin content of the epidermis may reduce the amount of light transmitted through the skin. Previous studies examining the impact of skin tone on photoplethysmographic HR measurement accuracy have produced mixed results; however, none measured epidermal melanin content.

Purpose: To determine whether objectively measured skin pigmentation influences the accuracy of photoplethysmographic HR measurement during rest, exercise, and recovery.

Methods: Skin pigmentation was quantified from colorimeter measures using Individual Typology Angle (ITA°), a strong correlate of epidermal melanin, in 28 healthy adults (White, n = 16; Black, n = 10; Middle Eastern, n = 1; Persian, n = 1). Chest-worn HR (criterion; Polar H10) and photoplethysmographic HR from three devices (Apple Watch Series 8, Garmin vivosmart 5, SlateSafety BAND V2) were recorded continuously during rest, cycling, and recovery; HR data were averaged into 30-s epochs for analysis. A linear mixed-effects model determined whether ITA° influenced mean absolute error of HR (MAE_HR) for each device.

Results: ITA° predicted MAE_HR for SlateSafety (β =  - 0.011, P = 0.001), but not for Apple (P = 0.62) or Garmin (P = 0.29). Missing data were disproportionately attributed to participants with dark skin (ITA° < 10°; 36%) for Apple (50%) and SlateSafety (85%) devices; outliers were disproportionately observed among participants with dark skin for all devices (56%-62%).

Conclusion: Data quality was negatively impacted by darker skin pigmentation for all tested devices, but skin pigmentation (ITA°) only increased photoplethysmographic HR measurement error to a small degree (~ 1 bpm) for the SlateSafety device.