A Preliminary Concurrent Validation of a Portable Markerless Motion Capture System Against Three-Dimensional Gait...

Three-dimensional gait analysis (3DGA) in patients undergoing total hip and knee arthroplasty (THA and TKA, respectively) is impractical for widespread clinical use. This study evaluated the Azure Kinect, combined with the Motognosis AMSA software (AK-M), for the assessment of selected clinically relevant gait parameters in THA and TKA patients compared with reference 3DGA. Twenty participants (10 THA, 10 TKA) were simultaneously assessed using AK-M and Qualisys 3DGA while walking along an 8 m walkway. Discrete peaks of sixteen bilateral gait parameters were compared using Bland-Altman mean bias and 95% limits of agreement (LOAs), with ICC(3,1) as a complementary measure of relative consistency. Maximum hip kinematics and knee extension demonstrated low mean bias. In contrast, maximum knee flexion, knee range of motion, and step length demonstrated larger mean bias and wider LOAs. Cadence and gait speed demonstrated the lowest bias and narrowest LOAs. ICC(3,1) values ranged from 0.84 to 0.95. AK-M showed parameter-specific agreement with 3DGA, supporting preliminary concurrent validity for simplified assessment of selected discrete gait outcomes. Wider LOAs for other parameters identify areas requiring further technical refinement and clinical interpretation. These findings apply to the combined arthroplasty cohort and should not be interpreted as separate THA and TKA cohorts.