In-clinic versus at-home gait kinematics in adults with and without stroke: Associations across environments

Importance Stroke remains a leading cause of mortality and long-term disability in the United States. Gait dysfunction following stroke can limit mobility, participation, and quality of life, even among individuals deemed capable of community ambulation. Gait is commonly assessed in clinical settings under the assumption that it reflects real-world gait; however, empirical support for this assumption remains limited. Objective To determine whether gait kinematics measured in a clinical setting reflect gait kinematics measured in real-world environments. Design Smartphone or tablet video recordings of participants walking in both clinic and home environments were collected and processed using pose estimation and MATLAB-based analysis to extract gait kinematics. Data from 43 participants (18 with history of stroke, 25 without history of stroke), were analyzed. Measures Gait kinematic variables derived from video-based pose estimation, including step length asymmetry (SLA), step time asymmetry (STA), and gait speed (GS). Results Spearman correlation analyses revealed weak-to-strong associations between in-clinic and at-home gait metrics with moderate-to-strong correlations among participants with history of stroke (SLA: 𝜌 = 0.455, p = 0.058; STA: 𝜌 = 0.519, p = 0.027; GS 𝜌 = 0.752, p=<0.001) and weak-to-moderate correlations among participants without history of stroke (SLA: 𝜌 = 0.116, p = 0.580; STA: 𝜌 = 0.406, p = 0.044; GS: 𝜌 = 0.541, p = 0.005). Wilcoxon signed-rank analyses indicated no statistically significant differences between most gait metrics between settings (α < 0.05), except STA among participants without history of stroke (p = 0.01). Conclusions These findings support the feasibility of using smartphone- or tablet-based video with pose estimation to assess real-world gait. Additionally, they suggest that clinic- and home-based gait are related but not identical. While several gait parameters demonstrated moderate-to-strong correlations between environments, others exhibited weak associations. Together, these findings indicate that clinic-based assessments may provide useful information regarding certain aspects of real-world gait but may not fully represent gait in the home environment.