The bit doesn't fit: Evaluation of a commercial activity-tracker at slower walking speeds

Abstract

Accelerometer-based commercial activity trackers are a low-cost and convenient method for monitoring and assessing health measures such as gait. However, the accuracy of these activity trackers in slow walking conditions on a minute-by-minute basis is largely unknown. In this study, the accuracy of a hip-worn commercial activity tracker (FitBit Ultra) was examined through step counts. Accuracy was evaluated through four minute trials of treadmill walking at speeds representative of older adults (0.9, 1.1, and 1.3 m/s). Minute-by-minute step count was extracted through the FitBit API and compared it to observer counted steps through video recordings. Results highlighted a significant over-reporting of steps at the highest speed, and a significant under-reporting of steps at the slowest speed, with the FitBit Ultra failing to count steps for one or more minutes at the slowest speed for 11 participants. This study highlights problems with using the FitBit Ultra by slow-walking populations, and recommends that researchers and clinicians should carefully consider the trade-off between accuracy and convenience when using commercial activity trackers with slow-walking populations.