Markerless motion capture is increasingly used in pediatric neurorehabilitation, yet its ability to detect patient-specific gait abnormalities in small and heterogeneous cohorts remains unclear. This study evaluated a smartphone-based markerless workflow (OpenCap integrated with OpenSim) as a clinical assessment tool to support individualized planning in the context of robot-assisted gait rehabilitation (RAGT) by characterizing individualized gait deviations in four pediatric patients with neurological gait disorders, referenced against normative data from 30 healthy individuals. Sagittal hip, knee, and ankle kinematics were extracted, normalized, and converted into gait-cycle–dependent Z-scores. Group-level comparisons using one-sample Statistical Parametric Mapping (SPM) revealed no significant deviations between patient-group means and normative trajectories (p ≥ 0.05). In contrast, individualized deviation profiling—including Z-score heatmaps, phase-wise Z-score analysis, and per-patient kinematic overlays—identified distinct, clinically meaningful abnormalities in every patient, such as excessive swing-phase hip and knee flexion, mid-stance knee extension deficits, reduced terminal-stance hip extension, and markedly diminished late-stance ankle plantarflexion and push-off. Several deviations exceeded |2–5| SD from the normative dataset, indicating substantial impairments that were obscured by group averaging. These individualized patterns were consistent with each patient’s clinical presentation and could be interpreted in relation to modifiable gait features that are commonly considered during planning and phase-specific adjustment of robot-assisted gait rehabilitation, rather than serving as direct evidence of therapeutic efficacy. Overall, the findings demonstrate that smartphone-based markerless motion capture enables sensitive, individualized gait assessment even when group-level statistics remain nonsignificant, supporting its use as an exploratory, decision-support framework rather than as an outcome measure of RAGT.
Yu Min (Fri,) studied this question.