ACCURACY OF HEALTH METRICS FROM WRIST-WORN FITNESS DEVICES DURING EXERCISE WITHIN DIVERSE COHORTS
Cooper R. Neeble, Cory M. Scott. Roanoke College, Salem, VA.
BACKGROUND: The ability to measure health metrics from commercially available wrist-worn fitness devices may provide valuable information for epidemiological and physiological research pertaining to energy expenditure, exercise intensity, and heart rate. However, despite the advancements in technology, significant error may still arise within the metrics being collected. The purpose of this study was to determine if skin tone and hand-dominance impact the accuracy of heart rate (HR) and energy expenditure (EE) in wrist-worn devices utilizing photoplethysmography. METHODS: Twelve volunteers (12 male, age 24+5.9 years) of diverse skin-tone completed a single day laboratory trial comprising of two 15-minute progressive exercise protocols using a treadmill and a Monark 828e cycle ergometer. Two smart-watch brands, Apple Series 3 (AS3) and Fitbit Versa 2 (FV2) were selected from their commercial availability at the time of this study. Participants wore the same model on identical locations of each wrist on their dominant (D) and non-dominant (ND) arm, with the AS3 placed near the ulnar head, and the FV2 placed 2-3 finger widths above the ulnar head. Oxygen consumption and electrocardiography were assessed utilizing a metabolic cart (MGC Diagnostics Ultima™ CardiO2®), and were used as the gold standard to compare the values of HR and EE derived from the smart watches. RESULTS: No significant differences were found between skin tone and the percent error in HR (AS3_D: 0.73+0.63, p=0.80; AS3_ND: 0.19+0.19, p =0.57; FV3_D: 4.01+8.87, p=0.61; FV3_ND: 4.87+8.93, p=0.25) and percent error in EE (AS3_D: 3.71+34.21, p=0.62; AS3_ND: 0.11+25.40, p=0.29; FV3_D: 0.28+28.60, p=0.66; FV3_ND: 1.01+25.75, p=0.26) for all watches. However, significant differences were found when comparing hand dominance and total measured EE within brands (AS3_D vs. AS3_ND: p<0.001; FV2_D vs. FV2_ND: p<0.001), but not with HR (AS3_D vs. AS3_ND: p=0.78; FV2_D vs. FV2_ND: p=0.41). CONCLUSIONS: All devices accurately measured HR within the acceptable error of 5% in our diverse population. However, EE estimates were up to 66% off from the reference value and were significantly different between identical brands. Individuals and clinicians should be aware of the strengths and limitations of wrist-worn devices and how it may impact exercise prescription and daily caloric intake.
Neeble, CR and Scott, CM
"ACCURACY OF HEALTH METRICS FROM WRIST-WORN FITNESS DEVICES DURING EXERCISE WITHIN DIVERSE COHORTS,"
International Journal of Exercise Science: Conference Proceedings: Vol. 16:
2, Article 323.
Available at: https://digitalcommons.wku.edu/ijesab/vol16/iss2/323