In this project, we developed simulation methods that can be used to help understand the underlying mechanics of movement disorders in children with cerebral palsy, and to aid treatment planning. Children with cerebral palsy suffer from multiple neuro-musculoskeletal deficits like decreased muscle selectivity, increased spinal reflex, increased muscle tone and skeletal deformities that cause gait deficits. Moreover, there is a big diversity in the presence and intensity of each deficit in different affected children, making it hard to study the underlying mechanics of individual deficits only using experiments. Indeed, the success rates of surgeries in children with cerebral palsy have not improved in the last 15 years. Recently developed state-of-the-art musculoskeletal modelling and simulations can help us understand the role of individual deficits and their interaction on gait deficits as we can develop physics-based models of each deficit and study their effect on movement in isolation and in presence of specific deficits. The same simulations can then be used to study the effect of addressing deficits in isolation and/or in combination to aid treatment planning. In this project, we improved the accuracy of the musculoskeletal model personalization that will improve the accuracy what we learn from musculoskeletal modelling and simulation studies and the accuracy of proposed treatment planning. Additionally, this project contributed towards making the state-of-the-art musculoskeletal modelling and simulations platform easy to use. An open source software called PredSim was created that can be used not only for studying children with cerebral palsy, but also human gait and running, disorders like stroke, Duchenne muscular dystrophy, walking with prosthesis, and even animal and bird locomotion.
Additionally, in this project, we also developed tools to study certain underlying mechanics in children with cerebral palsy that cannot be studied using PredSim. Sensory signals and motor commands are corrupted by errors called sensorimotor noise. A heightened gastrocnemius activity at heel-strike has been observed in children with cerebral palsy, and it has been theorized that the reason for this high gastrocnemius activity could be the presence of high sensorimotor noise and/or the interaction of sensorimotor noise with other deficits in children with cerebral palsy. Since PredSim does not account for sensorimotor noise, it is not able to predict the observed heightened gastrocnemius activity. Accounting for sensorimotor noise along with a complex control policy in simulations is computationally very expensive, and for that reason, most simulation generating platforms do not account for it. This limitation was recently addressed by prof. dr. Friedl De Groote. They developed a simulation platform that could generate simulations of walking while accounting for sensorimotor noise. This platform, however, used a simple stick figure human model without any feet or muscles. To study the role of sensorimotor noise, and its interaction with other deficits during gastrocnemius activity at heel-strike, the fellowship recipient added feet and muscles to the simulation platform that accounts for sensorimotor noise. These developments will enable researchers to study isolated an interaction effects of sensorimotor noise in children with cerebral palsy.
