Accuracy Of Research And Consumer Physical Activity Monitors In Estimating Energy Expenditure

Abstract

The use of accelerometers to estimate physical activity energy expenditure (EE) is common in a research setting. Recently, accelerometry-based physical activity monitors have been introduced for consumers but there are few data comparing EE estimation accuracy for these consumer devices compared to those used in research. PURPOSE: To determine the accuracy of EE estimates from a variety of physical activity monitoring devices during activities of daily living. METHODS: Nineteen adults (10 male, 9 female), mass: 75.14 (17.1) kg, BMI: 25.07(4.6) kg/m∧2, completed a four hour stay in a room calorimeter. Participants wore three physical activity monitoring devices used in research: Actical, Actigraph, were worn at the hip, and the IDEEA sensors were secured to the proper limb and chest positions. In addition, participants wore two consumer devices: Philips DirectLife and Fitbit. Each individual performed a series of randomly assigned postures/activities including supine, seated (quietly and using a computer), standing, walking, stepping, cycling, sweeping, as well as a period of self-selected activities. We used the device manufacturer's software to estimate EE and compared these estimates to the measured EE. The time resolution for DirectLife was one hour, and the epoch length of the Actigraph was one second, summed by the hour. Therefore, we were able to extract from these devices a 3hours of data, and the remaining devices had a resolution of between one and five minutes which allowed us to compare EE during a 3.5 hour period. RESULTS: The total 3 hour EE was 455(18) kcal and the total 3.5 hour EE was 503(18) kcal (mean(SE)). The DirecLlife (449(13) kcal) and IDEEA (445(23) kcal) estimates of EE were similar to the measured EE. The Actigraph, Fitbit and Actical devices significantly underestimated EE (339 (19) kcal, 363(18) kcal and 383(17) kcal,respectively). Root mean square errors were 62.1(14%), 88.2(18%), 122.2(27%), 130.1(26%), and 143.2(28%) for DirectLife, IDEEA, Actigraph, Actical and Fitbit respectively. CONCLUSIONS: The research and consumer physical activity monitors tested have a wide range of accuracy when estimating EE. Given the similar hardware of these devices, these results suggest that the algorithms used to estimate EE are primarily responsible for their accuracy.