An Acceleration-Based Nonlinear Time-Series Analysis of Effects of Robotic Walkers on Gait Dynamics During Assisted Walking

摘要

This article aimed to identify the effects of a robotic walker on gait dynamics during assisted walking via the analysis of acceleration data. Walking experiments were conducted with 11 healthy young subjects under different conditions, including normal, rollator-assisted, and robotic-walker-assisted walking. We collected acceleration data from the trunk and pelvis to extract measures for evaluating the properties of gait dynamics, including periodicity, complexity, and determinism. These measures include maximum Lyapunov exponent (MLE), multiscale entropy (MSE), multiscale permutation entropy (MPE), correlation dimension (CD), and percentage of determinism (%DET, from recurrence plot). Significant differences in the complex measures, MSE and MPE between normal and walker-assisted walking for trunk acceleration data in anteroposterior (AP) and mediolateral (ML) directions were observed. The determinism measures, CD and %DET, demonstrated significant differences between normal and robotic-walker-assisted walking for trunk acceleration data in the AP direction, while significant differences in the periodicity measures and MLE were not observed between the three conditions. These results imply mostly unchanged local dynamic stability, reduced deterministic patterns, and increased complexity in the gait dynamics of healthy young subjects, especially for the upper body. The robotic walker’s assistance is considered appropriate because of no significant difference between rollator-assisted and robotic-walker-assisted walking. This study provides new insights into the biomechanical effects of the robotic walker on gait. It also suggests that the used methods can be applied to the assistance evaluation of walking assistive devices like lower-limb exoskeletons and assess the ability to resist perturbations with a robotic walker for rehabilitation applications such as perturbation-based gait training.