Abstract
Walking speed is primarily driven by the propulsion force generated during the late stance phase, but the translation of propulsion force to walking speed may differ across individuals. This study aimed to investigate inter-individual variability in the relationship between propulsion force and walking speed in patients with subacute stroke and identify the clinical characteristics associated with this variability. Twenty-five participants with subacute stroke performed walking trials at self-selected and highest possible speeds. We applied hierarchical linear modeling, which allows the estimation of both group-level effects and individual-level variations, to determine whether adding individual's regression coefficients for propulsion forces improved the model fit. The association between an individual's regression coefficient and their clinical characteristics was examined using Bayesian network analysis. The results showed meaningful inter-individual variability in the effect of the propulsion force on walking speed. A probabilistic association from the Trunk Impairment Scale to the regression coefficients of paretic propulsion force (r = - 0.42) and another from the regression coefficients of non-paretic propulsion force to paretic propulsion force (r = - 0.48) were observed. Individuals with reduced trunk function may tend to rely more on paretic propulsion to maintain walking speed. These patient-specific propulsion patterns may facilitate the development of effective rehabilitation strategies.