•  
  •  
 

EQUIVALENCE TESTING OF WEARABLE TECHNOLOGIES DURING TREADMILL WALKING

Abstract

BACKGROUND: Step-counting is a wide-spread feature of most contemporary technologies and can be used measure to step-based physical activity (PA) as steps/day or cadence (steps/min). The validation of wearable technologies is essential to their selection for various purposes and to interpreting their data outputs. Equivalence testing has recently gained popularity as a statistical tool to determine agreement between devices, as opposed to traditional null hypothesis testing of statistical differences. PURPOSE: To determine step-count criterion validity of wearable technologies during treadmill walking using equivalence testing. METHODS: Data for this analysis was collected as part of the CADENCE-Adults study, a laboratory-based, cross-sectional investigation. Two hundred sixty adults (52.7±18.9 years, BMI 25.6±3.7, 50% women) performed a series of 5-min treadmill bouts from 0.5 to 5 mph, in 0.5 mph increments. Participants wore select wearable technologies on their waist (Yamax Digiwalker SW200 [SW200], New Lifestyles NL1000 [NL], ActiGraph GT9X [AG] and ActiCal [AC]), thigh (activPal [AP]), and ankle (StepWatch [SW]). The criterion measure was directly observed hand-counted steps (verified using video recording). Equivalence testing plots were constructed to evaluate criterion validity, with agreement between devices and the criterion, based on whether a device’s mean error and 95% CI fell within a proposed equivalence zone (±0.2 SD of the criterion step count for each speed). RESULTS: The SW200 fell within the equivalence zone at 4.0, 4.5 and 5.0 mph. The NL fell within the equivalence zone at 4.0, 4.5 and 5.0 mph. The AG fell within the equivalence zone at 3.5 and 4.5 mph. The AC displayed no statistical equivalence at any of the speeds. The AP fell within the equivalence zone for all speeds between 1.5 and 4.0 mph. The SW fell within the equivalence zone for all speeds between 0.5 and 3.5mph. CONCLUSION: Our findings add to a body of evidence regarding the step- count validity of wearable technologies through the use of equivalence testing across a full range of treadmill walking speeds. No device fell within the equivalence zone for all speeds tested. Of the devices tested, the SW and AP performed the best, particularly at slow to normal speeds, but with increasing error at the fastest speed tested. FUNDING: NIH-NIA-5R01AG049024

This document is currently not available here.

Share

COinS