Development and Validation of a Framework for Predictive Simulation of Treadmill Gait

Abstract

Treadmill training is a common intervention to promote healthy walking function for individuals with pathological gait. However, because of the heterogeneity of many patient populations, determining how an individual will respond to new treadmill protocols may require extensive trial and error, causing increased patient fatigue. The purpose of this study was to develop and validate a framework for predictive simulation of treadmill gait, which may be used in the design of treadmill training protocols. This was accomplished through three steps: predict motion of a simple model of a block relative to a treadmill, create a predictive framework to estimate gait with a two-dimensional (2D) lower limb musculoskeletal model on a treadmill, and validate the framework by comparing predicted kinematics, kinetics, and spatiotemporal parameters across three belts speeds and between speed-matched overground and treadmill predictive simulations. Predicted states and ground reaction forces for the block-treadmill model were consistent with rigid body dynamics, and lessons learned regarding ground contact model and treadmill motion definition were applied to the gait model. Treadmill simulations at 0.7, 1.2, and 1.8 m/s belt speeds resulted in predicted sagittal plane joint angles, ground reaction forces, step length, and step time that closely matched experimental data at similar speeds. Predicted speed-matched overground and treadmill simulations resulted in small root-mean-square error (RMSE) values within standard deviations for healthy gait. These results suggest that this predictive simulation framework is valid and can be used to estimate gait adaptations to various treadmill training protocols.