Kinematic modularity analysis of walking-based and in-place tasks

Abstract

Unlike the muscle synergies, the modularity of the kinematic synergies during different tasks has not been studied comprehensively. The objective of this study was to derive the shared and task specific kinematic modules for two different movement modes of walking-based and in-place tasks. Kinematic modules were extracted by applying the NNMF technique on the half-wave-rectified kinematic data of six subjects, taken from the CAMS-Knee dataset. A novel clustering methodology, using three steps of k-means, was employed to identify the shared and task-specific kinematic synergies. The results were analyzed for a variety of combined global and local variance accounted for (VAF) criteria. Assuming the VAF global and local to be 0.95 and 0.85, respectively, there were four inclusive shared modules for all walking-based tasks, and two for all in-place tasks, which were activated with different time profiles. In particular, a single module was activated at the beginning of the descending and the end of the ascending in-place tasks. It was concluded that the kinematic maneuvers observed in different tasks could be reconstructed by appropriate activation of a limited number of kinematic modulus.