Evaluation of four sensor locations for physical activity assessment

Abstract

Direct measurements of physical activity (PA) obtained with inertial measurement units (IMUs) secured to the upper arms and trunk of 36 registered nurses working a full shift were compared to measurements obtained with a commercially-available PA monitor (ActiGraph wGT3X-BT) worn at the waist. Raw accelerations from each device were summarized into PA counts/min and metabolic equivalent (METs) categories using standard definitions. Differences between measurements were examined using repeated measures one-way analyses of variance (ANOVA) and agreement was assessed using Bland-Altman plots. Statistically significant differences were observed between all sensor locations for all PA summary metrics except for between the left and right arm for percentages of work time in the light and moderate counts/min categories. Bland-Altman plots suggested limited agreement between measurements obtained with the IMUs and measurements obtained with the wGT3X-BT waist-worn PA monitor. Results indicate that PA measurements vary substantially based on sensor location.