Smartphone-based motion capture for gait quantification and symmetry analysis in moderate-stage stroke patients

Moderate-stage stroke patients (Brunnstrom recovery stages for the lower-limb IV - V) often retain subtle gait impairments that conventional assessments may overlook, hindering rehabilitation. While 3D gait analysis is precise, its cost and impracticality limit routine use. This cross-sectional pilot study evaluated the potential of a low-cost, dual-smartphone markerless motion capture system (OpenCap) for efficient and clinically useful gait quantification and symmetry analysis in this population.

Gait data were collected from 17 stroke patients and 20 healthy controls using OpenCap. Spatiotemporal parameters, symmetry indices, and sagittal-plane hip, knee, and ankle kinematics were computed. Comparisons between groups and between paretic/non-paretic limbs were performed using ANCOVA adjusting for sex, height, weight, and gait speed, with Bonferroni correction. Within-session reliability was calculated using intraclass correlation coefficients (ICC) from multiple trials per subject.

Stroke patients showed significantly longer stance and swing times, reduced stride length, and lower peak hip and knee flexion on the paretic side compared to controls (all p < 0.05, Cohen's d = 0.9-2.0). Asymmetry indices for swing time, hip flexion, knee flexion, and ankle dorsiflexion were significantly higher in stroke patients (p < 0.05, d = 0.7-1.7). Within-session reliability for joint angles ranged from moderate to excellent.

A two-smartphone markerless pipeline detected group-level gait differences and asymmetries in moderate-stage stroke patients, supporting its potential for accessible, clinic-friendly assessment and remote monitoring.