Assessing the Validity and Reliability of a Markerless Motion Capture System for Sagittal-Plane Gait Range of Motion

Optical three-dimensional (3D) motion capture systems and inertial measurement unit (IMU)-based systems provide accurate kinematic measurements, yet both remain costly and require complex setups. Artificial intelligence (AI)-driven markerless systems that operate with a single smartphone offer a practical alternative. In this study, we evaluated the validity and reliability of a markerless motion capture system (SPLYZA Motion; SPLYZA Inc., Shizuoka, Japan) for sagittal-plane gait analysis relative to an IMU-based 3D system (Ultium Motion; Noraxon Inc., Scottsdale, AZ, USA).

Twenty healthy adults walked at a self-selected pace, and hip, knee, and ankle joint angles were recorded simultaneously using both systems. Joint angles were time-normalized, and waveform characteristics and peak values were compared. Associations were assessed using Spearman's rank correlation (ρ), agreement was examined using intraclass correlation [ICC(3,1)] with 95% confidence intervals (CIs), and proportional and fixed bias were evaluated using linear regression and Wilcoxon signed-rank tests. Agreement was further assessed using Bland-Altman analysis.

Significant positive correlations were observed for the hip (ρ = 0.913, p < 0.01), knee (ρ = 0.833, p < 0.01), and ankle (ρ = 0.721, p < 0.01). ICCs demonstrated excellent agreement for the hip (0.914; 95% CI: 0.906-0.920) and knee (0.934; 95% CI: 0.929-0.940) and good agreement for the ankle (0.723; 95% CI: 0.701-0.743). Root-mean-square error (RMSE) values (mean ± SD) were 7.73 ± 3.32° (hip), 7.48 ± 3.40° (knee), and 6.38 ± 2.44° (ankle). Bland-Altman plots indicated small biases for the hip and knee, with wider limits for the ankle near end-range positions.

SPLYZA Motion showed accuracy and reliability comparable to Ultium Motion within clinically acceptable limits. Minor ankle-related biases were small enough to avoid affecting clinical interpretation. Given its low cost and portability, this system provides a practical option for gait assessment in routine clinical and research environments.

Minimal setup AI-based markerless motion capture enables a feasible clinical gait analysis.