Clinical Validation of Smartphone-Based Markerless Motion Capture for Dynamic Cervical Alignment During Gait in Patients With Cervical Myelopathy

Study Design: Prospective validation study of smartphone-based gait analysis technology. Objective: To establish the clinical feasibility, test-retest reliability, and criterion-related validity of a smartphone-based Three-Dimensional Pose Tracker for Gait Test (TDPT-GT) in patients with cervical myelopathy. Summary of Background Data: Objective dynamic assessment of cervical spine alignment during functional activities has been limited by the requirement for expensive motion capture systems. Recent advances in artificial intelligence enable markerless motion capture using standard smartphones, but clinical validation in cervical myelopathy patients is lacking. Methods: Ninety-two consecutive patients with cervical spondylotic myelopathy or ossification of the posterior longitudinal ligament underwent TDPT-GT assessment. The system uses deep learning algorithms to track body keypoints during circular walking. Two measurements were performed for reliability analysis. Clinical correlates included the Japanese Orthopaedic Association (JOA) score, the modified Rankin Scale (mRS), the instrumented Timed Up and Go (iTUG), and static radiographic parameters (C2–7 SVA, C2–7 angle, T1 slope). Results: All 92 patients completed the assessment (100% feasibility). Test-retest reliability for TDP neck angle was good [ICC(3,1)=0.851, 95% CI: 0.780–0.900], with SEM of 1.82° and MDC 95 of 5.06°. Bland-Altman analysis showed minimal bias (−0.10°) with 95% limits of agreement from −5.16° to 4.96°. TDP neck angle correlated significantly with JOA score ( ρ =–0.379, P <0.001), C2–7 SVA ( ρ =0.401, P <0.001), and iTUG time ( ρ =0.269, P =0.010). The correlation between dynamic and static cervical parameters supports the construct validity of smartphone-based assessment. Conclusion: TDPT-GT demonstrates good test-retest reliability and meaningful correlations with established clinical measures in cervical myelopathy patients. This accessible technology enables standardized dynamic cervical alignment assessment in routine clinical practice. Levels of Evidence:​​​ Level II—diagnostic study.