[A study on the evaluation of motor function in cervical spondylotic myelopathy using gait and plantar pressure parameters based on visual recognition].

Objective: To investigate gait differences between patients with cervical spondylotic myelopathy (CSM) and healthy individuals, and to further analyze gait characteristics across different disease severities and gender subgroups, utilizing vision-based gait analysis and plantar pressure technology. Methods: This is a cross-sectional study. From April to August 2024, 147 CSM patients and 154 asymptomatic volunteers were recruited at the Department of Spine Surgery, the Second Affiliated Hospital, Naval Medical University. There were 168 males and 133 females, aged (54.1±10.4)years (range:28 to 76 years). Propensity score matching using the nearest neighbor method (caliper value 0.02) was performed with gender, age, and body mass index as matching variables. Gait parameters were obtained by analyzing walking videos captured via a smartphone computer vision algorithm (OpenPose framework). Plantar pressure distribution was measured using wireless sensor insoles (sampling rate 50 to 100 Hz). Intergroup comparisons were conducted using independent sample t-tests, and multi-group comparisons used analysis of variance. The diagnostic cut-off values for gait-related indicators were determined via receiver operating characteristic (ROC) curves, area under the curve (AUC), and the Youden index. Results: After PSM, a total of 102 matched pairs were obtained. There were no statistically significant differences in the general data between the CSM group and the healthy control group after matching (all P>0.05). At self-selected walking speed, the CSM group exhibited slower gait speed ((0.63±0.19) m/s vs. (0.83±0.26) m/s, t=9.310, P<0.01) and cadence ((39.38±16.99) steps/min vs. (46.12±13.63) steps/min, t=4.595, P<0.01), shorter stride length ((0.43±0.09) m vs. (0.52±0.12) m, t=9.434, P<0.01) and step length ((0.85±0.19) m vs. (1.01±0.23) m, t=8.998, P<0.01), prolonged gait cycle ((1.23±0.17) s vs. (1.14±0.17) s, t=-3.715, P=0.001), increased proportions of stance phase ((66.96±5.40)% vs. (54.11±9.57)%, t=16.952, P<0.01) and double support phase ((33.13±10.74)% vs. (21.78±8.46)%, t=16.699, P<0.01), and a decreased proportion of swing phase ((33.03±5.40)% vs. (45.73±9.41)%, t=11.701, P<0.01). Multiple kinematic parameters (hip flexion angle, ankle dorsiflexion angle) were significantly lower in the CSM group (all P<0.01). Plantar medial pressure was increased ((44.62±15.81)% vs. (39.07±13.11)%, t=-3.471, P=0.016), while lateral pressure was decreased ((55.38±15.81)% vs. (60.93±13.11)%, t=2.721, P=0.019). Among these parameters, the stance phase proportion demonstrated the highest diagnostic accuracy (AUC=0.944, sensitivity=88.9%, specificity=89.2%), with an optimal cut-off value of 64.75%. Conclusions: Significant differences in gait speed, stride length, gait cycle, hip flexion angle, ankle dorsiflexion angle,plantar medial and lateral pressure exist between CSM patients and healthy controls. A stance phase proportion greater than 64.75% can serve as a primary gait parameter for assisting in the diagnosis of CSM.