Investigating the neuromechanical control of healthy gait modulation and pathological gaits observed in cerebral palsy using neuromuscular simulations

Locomotion is based on a sophisticated interaction among the environment, the musculoskeletal system, the spinal cord, and the brain locomotor areas. Quality of life is strongly related to the proper capability of this movement. However, many pathologies, like cerebral palsy (CP), lead to motor impairments that limit locomotor capabilities. The proper treatment choice depends on the relationship between biomechanical or neural impairments and the observed gait deviations. Predictive neuromuscular simulations can be powerful tools for testing how specific impairments influence gait deviations and the roles of different neural structures in controlling locomotion. This thesis has two main objectives to pursue using neuromuscular simulations: (1) investigating the roles of spinal reflexes and central pattern generator (CPG) circuits in the control and modulation of healthy human locomotion, (2) investigating the gait deviations associated with biomechanical and neural impairments commonly observed in CP patients and evaluating the effects of muscle-tendon surgeries. First, we perform these investigations using purely reflex-based controllers guided by finite state-machines adapted from previous studies. Our results suggest that reflex gains and thresholds of hip muscles and ankle plantarflexors can be modulated by the descending mechanisms to achieve different gait behaviors in healthy human locomotion. Furthermore, pathological behaviors like heel and toe walking can be related to both biomechanical and neural impairments. More specifically, contracture or hyperreflexia applied to plantarfleor can reproduce toe walking, whereas partial heel walking was achieved through plantarflexor biomechanical or neural weaknesses. Then, we developed a new neuromuscular controller composed of a CPG network with five locomotor synergies and a more sophisticated reflex network built using leaky integrator neurons without relying on a finite state-machine. This controller is used to explore the two investigation objectives of this thesis. A vast range of walking speeds can be reproduced with this model, and results suggested that CPGs could be an important component in generating rhythmic locomotion by preventing or enabling reflexes to affect gait in specific gait cycle phases. Furthermore, CPG's frequency seems to be crucial in modulating step duration. Concerning gait pathologies, we use the newly developed controller to investigate the effect of biomechanical contractures and neural spasticity affecting the hamstring muscle in generating crouch gait behaviors. Results suggest that crouch gait could be generated with moderate or severe levels of contracture and spasticity or a combination of these. Then, we simulate the benefit of different levels of hamstring lengthening surgery for several combinations of biomechanical and neural impairments. Results suggest that a moderate level of surgery could improve gait deviations related to muscle contractures by decreasing the amount of passive force generated. The surgery appears much less effective when the gait deviation is primarily due to neural spasticity. Furthermore, severe surgeries may result in problems in generating meaningful tension in the hamstring muscle-tendon unit. Our work shows the great potentialities of predictive neuromuscular simulations in investigating open questions in motor control, the emergence of pathological behaviors, and the possible effects of clinical treatments.