Locomotion results from the interactions of highly nonlinear neural and biomechanical dynamics. Accordingly, understanding gait dynamics across behavioral conditions and individuals based on detailed modeling of the underlying neuromechanical system has proven difficult. Here, we develop a data-driven and generative modeling approach that recapitulates the dynamical features of gait behaviors to enable more holistic and interpretable characterizations and comparisons of gait dynamics. Specifically, gait dynamics of multiple individuals are predicted by a dynamical model that defines a common, low-dimensional, latent space to compare group and individual differences. We find that highly individualized dynamics - i.e., gait signatures - for healthy older adults and stroke survivors during treadmill walking are conserved across gait speed. Gait signatures further reveal individual differences in gait dynamics, even in individuals with similar functional deficits. Moreover, components of gait signatures can be biomechanically interpreted and manipulated to reveal their relationships to observed spatiotemporal joint coordination patterns. Lastly, the gait dynamics model can predict the time evolution of joint coordination based on an initial static posture. Our gait signatures framework thus provides a generalizable, holistic method for characterizing and predicting cyclic, dynamical motor behavior that may generalize across species, pathologies, and gait perturbations.

Background: A common perspective in post-stroke gait training is that walking at the fastest safe speed maximizes the quality of gait biomechanics, with limited detrimental effects on compensatory biomechanics and inter-limb asymmetry. This fastest is best perspective is highly relevant to treadmill training paradigms, as mass high-intensity stepping practice with high-quality biomechanics can improve walking function and reinforce desirable gait patterns. However, it is unclear if walking at the fastest safe speed maximizes the quality of (i.e., optimizes) post-stroke gait biomechanics across variables, individuals, and walking function levels, or if there exists a significant cost (i.e., benefit lost) of walking at the fastest speed when fastest is not optimal. Methods: Here, we determined if walking at the fastest speed optimized 16 biomechanical magnitude and inter-limb asymmetry variables, in 14 low- (n=7) and high-functioning (n=7) stroke survivors. Participants walked at six speeds ranging from their self-selected to fastest safe speed. To characterize the relative benefit of optimizing, rather than maximizing, gait speed for each variable, we compared the biomechanical cost (i.e., immediate speed-induced change versus the self-selected speed) of walking at the fastest versus the optimal speed. Finally, we used linear regression to characterize how each variable's quality changed with absolute speed. Results: Across speeds, 50% of magnitude and 17% of asymmetry variables were optimized at the fastest speed, but which variables were optimized differed between participants. Compared to walking at the optimal speed for each variable, the fastest speed elicited large biomechanical costs for some inter-limb asymmetry variables (difference in Cohen's d=0.1-0.9). Both low- and high-function subgroups exhibited significant positive correlations between walking speed and paretic-leg trailing limb angle, peak ankle moment, and peak hip and ankle power magnitudes (all p<0.001), though the magnitude of changes in some variables differed between groups. Changes in inter-limb asymmetry were highly variable, even within-groups. Conclusions: These results refine the perspective that fastest is best, showing that the training speeds that maximize gait quality may not be the fastest for all individuals and biomechanical variables. Individual-specific stroke gait quality metrics encompassing multiple biomechanical variables are needed to guide gait speed optimization for precision rehabilitation.

No abstract