Modeling Gait Adaptation During a Novel Walking Task

• Varun Ravichandar, Biomedical Engineering, University of Delaware

• Joshua Cashaback, Biomedical Engineering, University of Delaware

Stroke survivors often develop persistent step length asymmetries that reduce walking efficiency, increase fall risk, and limit community mobility, making restoration of gait symmetry a central goal in rehabilitation. Treadmill-based gait training can improve symmetry, but existing approaches typically train one limb at a time or rely on speed manipulations rather than giving walkers continuous information about the symmetry of their steps. To address this gap, we designed a training paradigm in which participants simultaneously adapt their left and right step lengths using a visual target that represents their step length symmetry. The goal of this project was to determine how individuals adapt their gait when receiving this novel form of visual feedback and to identify which model of motor adaptation best describes the observed behavior.

Participants walk on a treadmill while viewing real-time feedback about the symmetry between their left and right step lengths. They are instructed to adjust their steps to match target values that require an asymmetric step pattern. After data collection, behavioral data will be fit to three candidate motor adaptation models which use a state-space approach: one-state model with only a learning parameter, a one-state model with both learning and retention parameters, and a two-state model with distinct fast and slow learning and retention processes. Model performance will be compared using mean squared error, Akaike Information Criterion, Bayesian Information Criterion.

By identifying the most appropriate model for this novel symmetry-targeted treadmill task, this project aims to characterize gait adaptation when both limbs are trained together with explicit visual feedback. In the future, understanding these adaptation mechanisms may help guide the design of treadmill-based rehabilitation protocols that use targeted visual feedback to help individuals who have experienced stroke regain more symmetrical walking patterns and improve functional mobility.