| Sumario: | Wearable activity monitors (AM) measure steps, heart rate (HR), and energy expenditure (EE). A strong correlation (r = 0.93) has been found between predicted EE by AM and indirect calorimetry during short duration exercise (Kendall et al., 2019). It is unknown if predicted EE is accurate during long duration exercise. PURPOSE: To determine the accuracy of predicted EE by a wrist AM during short and long duration exercise and to determine differences in accuracy between moderate and high fit individuals. METHODS: Participants (n = 31) were recruited to the EMU Running Science Laboratory on three occasions. At visit one, a VO2max treadmill test was completed at a maintained self-selected pace, while grade increased by 2% every 2-minutes until volitional exhaustion. Additionally, running speed at 70-75% of VO2max was determined. Each VO2max was classified using ACSM's normative values and split evenly into moderate and high fit groups. Visit 2 was a 10- or 30-minute treadmill run at the speed found at the first visit, with visit 3 being the other timed run. Timed run order was randomized and counterbalanced. A wrist AM was worn to predict EE and a metabolic cart recorded measured EE continuously throughout the test. A Pearson correlation compared predicted and measured EE. Repeated measures ANOVAs determined the effect of duration and fitness level on EE estimates (p < 0.05). Bland-Altman plots showcased mean error and limits of agreement between measurement types for both runs. RESULTS: Twenty-five participants completed all three visits (15 males, aged 23.0 (5.0), VO2max 50.4 (7.0) mL/kg/min). A strong correlation was found between predicted EE and measured EE for both short (R = 0.860) and long (R = 0.785) duration (p < 0.001). Repeated measures ANOVA determined the interactive effect of measurement mode and fitness level was significant (p=0.006). Significant differences were found between measurement methods in the high fit group during short and long duration runs (p < 0.001; p = 0.003 respectively). CONCLUSION: Overall, there is a strong correlation between criterion and predictive measurements. However, there is an effect of fitness level on estimation accuracy that may be due to the EE algorithm. Future research should focus on the algorithm variables to determine what causes these inaccuracies across fitness levels.
|
|---|