Congruent Accuracy of Wrist-worn Activity Trackers during Controlled and Free-living Conditions

• Published in: International journal of exercise science Vol. 11; no. 7; p. 574
• Main Authors: Dondzila, Christopher J, Lewis, Colleen, Lopez, Justin R, Parker, Tonya
• Format: Journal Article
• Language: English
• Published: Bowling Green TopSCHOLAR 01-01-2018
• Subjects: Accuracy; Heart rate
• ISSN: 1939-795X

To examine activity tracker accuracy for measuring steps, energy expenditure, and heart rate in controlled and free-living conditions. Forty participants performed four, five-minute stages (walking: 53.7 m·min-1, 80.5 m·min-1; running: 134.1 m·min-1, 160.9 m·min-1) while wearing the Fitbit Charge HR (FB) and the Mio FUSE (MF) activity trackers. Measurements included steps, energy expenditure (kcals), and heart rate (beats·min-1). In addition to the FB and MF, participants wore the NL-1000 (NL) activity tracker during waking hours of the subsequent day. One way ANOVAs with Tukey’s post hoc analyses were performed to compare mean values for steps, kcals, and mean heart rate between the FB, MF, and criterion measures. Levels of agreement for heart rate with 95% confidence intervals were examined with Bland-Altman plots. Compared to criterion measures, the FB and MF underestimated steps and overestimated kcals at 53.7 m·min-1 (FB: 12.7% for steps, 89.2% for kcals; MF: 15.8% for steps, 44.9% for kcals, p<.001) and 80.5 m·min-1 (FB: 9.7% for steps, 69.9% for kcals; MF: 13.4% for steps, 32.0% for kcals, p<.001). During free-living conditions, the MF significantly underestimated steps by 30.0% (p<.05). Increasing exercise intensity is indicative of heightened accuracy for step detection and kcal estimation for the FB and MF, while decreasing heart rate accuracy for the FB. However, the MF performed poorly for estimating total daily activity.