Accuracy Of Energy Expenditure Estimates From The Bodymedia Sensewear® Pro 2 Armband

Abstract

The BodyMedia SenseWear® Pro 2 (SP2) Armband uses an array of sensors including accelerometers, heat flux, galvanic skin response, and skin temperature to identify bouts of activity, then employs specific algorithms to estimate energy expenditure (EE). This sensor array may provide improved responsiveness in EE estimation of activities not typically detected by waist mounted accelerometers. PURPOSE The purpose of this study is to test the accuracy of the BodyMedia equation for EE across a range of activities and to determine if the addition of heart rate (HR) would further improve the prediction of EE. METHODS Participants included 30 college age males (n = 13) and females (n = 17). Participants wore an SP2 and a HR monitor while completing a laboratory testing protocol consisting of 7 activities performed in 5-min stages with 1-min rest between stages. The activities performed were 1) standing arm swings (AS) 2) walking 3) walking w/AS 4) walking w/ load 5) walking w/grade 6) walking w/ grade and load 7) walking w/ grade, load, and AS. The protocol used a treadmill walking speed of 3.2mph, backpack load of 4.8kg, and grade of 5%. EE (kcal/min) was measured by open circuit indirect calorimetry (CART) and steady state EE values were compared with time synchronized EE estimates from the SP2 monitor. Repeated measures ANOVA was used to test for differences in estimated EE. Regression analyses were used to determine if the addition of HR would improve the EE estimation. RESULTS Correlations between CART and SP2 values were moderate to high for the locomotor activities (r = 0.77 to 0.88) but lower for the standing arm swings (r = 0.44). Repeated measures ANOVA showed no significant differences between CART and SP2 values for walking and walking w/ load but significant differences were noted for the remaining five activities involving grade or AS. The SP2 overestimated EE values for activities 1 and 3, and underestimated EE values for activities 5, 6, and 7. Effect sizes of the differences ranged from 0.39 to 1.32. Regression analyses showed a significant change in R2 from 0.60 to 0.65 with the inclusion of HR. CONCLUSION The SP2 monitor demonstrated a responsiveness to the increasing EE requirements of the activities but significant differences were noted with measured EE for activities involving grade and arm swings. The addition of HR significantly improved estimation of EE. Direct incorporation of HR into the pattern recognition software may be more effective.