Reliability and agreement between two wearable inertial sensor devices for measurement of arm activity during walking and running gait

Abstract

Study DesignThis is a validation study. BackgroundTracking limb movement with body worn sensors allows clinicians to measure limb dynamics to guide treatment for patients with movement disorders. The current gold standard, 3-dimensional optical motion capture, is costly, time-consuming, requires specific training, and is conducted in specialized laboratories. PurposeThe purpose of our study was to a compare consumer-grade inertial sensor to a laboratory-grade sensor to provide additional methods for capturing limb dynamics. MethodsThe participants wore an Apple Watch and a laboratory-grade Xsens sensor on each wrist during 3 conditions: walk, fast-walk, and run. Acceleration data were collected simultaneously on each device per wrist for all conditions. Intraclass correlation coefficients and Bland-Altman plots were calculated to measure intra-/interdevice reliability, evaluate bias, and limits of agreement. ResultsIntradevice ICCs showed good reliability during walk and fast-walk (0.79-0.87) and excellent reliability during run (0.94-0.97) conditions. Inter-device ICCs yielded moderate reliability during walk (0.52 ± 0.22) and excellent reliability in fast-walk and run (0.93 ± 0.02, 1.00 ± 0.01) conditions. Bland-Altman plots showed small biases with 90% or more of the data contained within the limits of agreement. DiscussionOur study demonstrates reliability and agreement between the two devices, suggesting that both can reliably capture upper extremity motion data during gait trials. ConclusionOur findings support further study of consumer-grade motion trackers to measure arm activity for clinical use. These devices are inexpensive, user-friendly, and allow for data collection outside of the laboratory.