Gait analysis provides objective metrics to evaluate mobility in populations such as individuals with knee osteoarthritis. However, in-lab assessments may not reflect real-world gait. Wearable inertial sensors offer a promising alternative, but few studies have directly compared concurrent gait measures from motion capture and wearable sensor-based gait analysis with free-living data collections. This study collected gait data preoperatively from 45 older adults with end-stage knee osteoarthritis using a concurrent collection of in-lab markerless motion capture and wearable sensors placed on the proximal shank, followed by up to seven days of continuous free-living wearable sensor recordings. Agreement between measurement types and effect size were calculated for concurrently available gait measurements, such as stride, stance, and swing times, as well as peak mediolateral angular velocity of the proximal shank segment. In-lab sensor and motion capture demonstrated excellent agreement, particularly for stride time (r = 0.96), while free-living sensor data captured slower, more variable gait with lower peak angular velocity (288 deg/s vs 254 deg/s). In-lab gait variables correlated more strongly with function (r = -0.45) and depressive symptoms (r = 0.43), whereas free-living peak angular velocity was weakly associated with pain (r = -0.31). Agreement between in-lab and free-living measures was lower (ICC > 0.67), though peak angular velocity retained moderate-to-good agreement (ICC = 0.75). These findings suggest that certain spatiotemporal gait metrics, particularly peak angular velocity, may be viable for use in extended free-living assessments. Across settings, peak angular velocity showed correlations with function comparable to gait speed, supporting its potential as a clinically relevant metric. The study supports the integration of wearable sensors into long-term gait monitoring for clinical populations. • Gait variables from in-lab sensor and motion capture show excellent agreement. • Lower levels of agreement between in-lab and free-living gait data. • Peak angular velocity shows promise as a native IMU measure for free-living analysis.
Ruder et al. (Fri,) studied this question.