Hybrid models of the neuromusculoskeletal system improve subject-specificity.

Muscle-actuated simulations of pathological gait have the capacity to identify muscle impairments and compensatory strategies, but the lack of subject-specific solutions prevents the prescription of personalized therapies. Conversely, electromyographic-driven models are limited to muscles for which data are available but can capture the true neural drive initiated by an individual subject. In order to improve subject-specificity and enforce physiological constraints on muscle activity, we propose a hybrid strategy for the optimization of subject-specific muscle patterns that involves forward dynamic simulation of whole body movement coupled with electromyographic-driven models of muscle subsets. In this paper we apply the hybrid approach to an example of post-stroke gait and demonstrate its unique ability to account for the unusual muscle activation patterns and muscle properties in patients with neuromuscular impairments.