Experimental analyses of human movement using motion capture techniques and electromyographic recordings (EMG) can qualitatively correlate observed movement patterns with the underlying muscle coordination. However, computer simulations using musculoskeletal models are needed to quantify the relationship between individual muscle actions and the overall movement mechanics. Since the human musculoskeletal system is redundant in actuation i.e, more muscles than degrees of freedom, most methods need to rely on some sort of optimization procedure to resolve muscle redundancy.

This dissertation essentially focused on the formulation of novel optimization based methods to answer two research questions of clinical significance 1) Determining muscle coordination during healthy and impaired movements and 2) Predicting how an impaired subject’s muscle and joint coordination could be optimally altered in order to achieve additional functionality. A novel simulation methodology which used muscle synergies derived from experimental EMG was developed to generate accurate tracking simulations of healthy and post-stroke walking. This approach allowed motor control strategies responsible for a certain range of movements to be captured in a synergy recombination model. Such a recombination model developed for post-stroke gait was used to simulate healthy gait data in order to study how stroke impaired motor control could theoretically compensate and recover healthy walking mechanics. The dissertation formulated an optimal control problem solved by direct collocation which sought alternative kinematic and muscle coordination strategies which enhanced specific functional characteristics of an upper extremity movement as a proof of concept.

The synergy driven simulations serve to assess muscle coordination of impaired subjects based on experimental motion data, and the optimal control approach serves to predict how the impaired kinematic and muscle coordination could be optimally altered to restore normal functionality. Thus when sequentially arranged, these simulations would form a pipeline which uses experimental motion data as an input and suggests rehabilitation strategies as the final output, which has potential for tremendous clinical utility.