Abstract
Consumer wrist-worn activity trackers are widely used to monitor physical activity and physiological responses during exercise, yet their accuracy across exercise intensities and physiological variables remains uncertain. PURPOSE: To evaluate the accuracy of wrist-wearable fitness trackers for heart rate (HR), oxygen saturation (SpO₂), energy expenditure (EE), and step count across exercise intensities. METHODS: Thirty (30) recreationally active adult males and females (n = 30; age = 22.8 ± 4.5 yr) completed four sessions consisting of informed consent, anthropometric and baseline measurements, and three treadmill exercise sessions. Exercise sessions included walking at 3 mph (0% incline), walking at 3 mph (12% incline), and jogging at 5 mph (0% incline) for 20 min each. All measurements were recorded every 5 minutes. HR was simultaneously recorded using a Polar heart rate sensor, step count was measured using a handheld tally counter, and SpO₂, was measured using a pulse oximeter. Participants wore Garmin Forerunner 745 [Garmin, USA], and Fitbit Sense 2 Advance Health Smartwatch. EE was estimated using the ACSM equations: VO2 (ml/kg/min) = (0.1 x speed (m/min)) + (1.8 x speed (m/min) x grade) + 3.5. Energy Expenditure (Kcal)= [VO2 (ml/kg/min) x Body Weight (kg) x Time (min) / 1000] x 5. RESULTS A repeated-measures ANOVA compared step count, HR, EE, and SpO₂ across three intensities: 3mph, 3mph at 12% incline, and 5mph. Step count from both devices provided significantly lower counts compared to the manual tally counter at 3mph (p< 0.001, η²= 0.26) and 3mph at 12% incline (p= 0.004, η²= 0.19), while no significant differences were observed at 5mph (p= 0.226, η²= 0.05). HR differed significantly between devices at 3mph (p< 0.001, η²= 0.27), 3mph at 12% incline (p= 0.002, η²= 0.19), and 5mph (p< 0.001, η²= 0.61), with significant device × time interactions occurring at both the 12% incline (p< 0.001) and 5mph intensities (p< 0.001). EE caloric estimation also varied significantly by device at 3mph (p= 0.003, η²= 0.27) and 5mph (p< 0.001, η²= 0.41), where Bonferroni post-hoc tests demonstrated that the Fitbit significantly overestimated calories compared to both the Garmin and the ACSM equation (p< 0.021). SpO₂ levels showed no significant differences between the pulse oximeter and the Garmin at 3mph (p= 0.501), 3mph at 12% incline (p= 0.79), or 5mph (p= 0.806) with significant device × time interactions occurring at 3mph (p= 0.03, η²= 0.15).CONCLUSION: Step count accuracy is speed-dependent: devices undercount at 3 mph and inclines but become reliable at 5 mph. Heart rate measurements are inconsistent across intensities and vary between devices over exercise duration. Fitbit overestimates energy expenditure compared with Garmin and the ACSM equation. Therefore, these devices are appropriate for oxygen saturation and higher-speed step monitoring but should be used cautiously for heart rate and caloric estimates due to brand-specific bias.
Recommended Citation
Azpeitia Aranda, Jabeth; Karabulut, Murat; Schlatter, Rebekah D.; Perez, Guillermo; and Karabulut, Ulku
(2026)
"Testing the Accuracy of Commercial Wrist-Wearable Activity Trackers at Different Exercise Intensities,"
International Journal of Exercise Science: Conference Proceedings: Vol. 2:
Iss.
18, Article 257.
Available at:
https://digitalcommons.wku.edu/ijesab/vol2/iss18/257