Steps to Facilitate the Use of Clinical Gait Analysis in Stroke Patients : Validation of a Single 2d Rgb Smartphone Video-Based System for Gait Analysis

Abstract

Introduction: Clinical gait analysis plays a central role in the rehabilitation of stroke patients. However, practical and technical challenges limit their use in clinical settings. This study aimed to validate SMARTGAIT, a deep learning-based gait analysis system that addresses these limitations.Methods: Eight stroke patients took part in the study at the Human Performance Research Centre of the University of Konstanz. Gait measurements were taken using both the marker-based Vicon motion capture system and the single- smartphone based SMARTGAIT system. We evaluated the agreement for knee, hip and ankle joint angle kinematics and spatiotemporal gait parameters between the two systems.Results: The results demonstrated mostly high levels of agreement between the two systems, with Pearson correlations of 3 0.79 for all lower body angle kinematics in the sagittal plane and 3 0.71 in the frontal plane. RMSE values were ≤ 4.6°. The intraclass correlation coefficients for all derived gait parameters showed good to excellent reliability.Conclusion: The results suggest that SMARTGAIT is a promising tool for gait analysis in stroke, particularly for quantifying gait characteristics in the sagittal plane, most relevant for clinical gait analysis. However, further analyses are required to validate SMARTGAIT in larger samples and its transferability to different types of pathological gait. In conclusion, a single smartphone recording (monocular 2D RGB camera) could make gait analysis more accessible in clinical settings, potentially simplifying the process and making it more feasible for therapists and doctors to use in their day-to-day practice.