Concurrent validity of IMU and phone-based markerless systems for lower-limb kinematics during cognitively-challenging landing tasks.

Abstract

Alternative technologies, such as IMU and low-cost markerless systems, may overcome the drawbacks of optoelectronic marker-based motion capture systems (OMS) in sports non-contact injury risk screening, but the precision of collected kinematic data must be validated in comparison to OMS. This study assessed the concurrent validity of the Xsens IMU and phone-based OpenCap systems for lower-limb kinematics during cognitively-challenging landing tasks. Thirty competitive athletes (13 females, 17 males) performed unplanned jump-land-jump tasks towards lateral secondary directions while kinematics was simultaneously recorded with OMS, Xsens, and OpenCap. The agreement of lower limb joint discrete (initial contact and peak) values was assessed using Bland-Altman plots. Kinematic waveforms validity and similarity were evaluated through RMSE, normalized RMSE (NRMSE), and coefficient of multiple correlation (CMC). All systems were also compared using statistical parametric mapping (SPM) ANOVA. Time series exhibited very strong similarity (CMC > 0.85) on the sagittal plane for both systems. Highly variable validity and agreement were found based on the joint and plane considered. The lowest error and bias were found for knee flexion (NRMSE ≤ 10%), while the hip rotation demonstrated the lowest agreement for both systems. SPM reported significantly different clusters across the contact phase between the systems for most kinematic variables. The findings corroborate that IMU- and phone-based systems generate sagittal joint kinematic waveforms that are comparable in shape with respect to OMS, although magnitude differences were observed for hip flexion. However, the validity of kinematics in the transverse and frontal plane (knee data not available in OpenCap) was limited, as variability and systematic errors must be acknowledged.