Pain, injuries or diseases might affect how we (are able to) coordinate movement. Therefore, an in-depth understanding of motor control, human movement dynamics and how pathologies affect movement coordination is essential to inform clinical practice that aims to improve the quality of movement in patients and therewith their quality of life.
Musculoskeletal models allow for efficient simulations of human movement dynamics to predict the forces in muscles and joints in a non-invasive manner. However, assumptions on motor control are required to solve Bernstein’s problem of muscle redundancy: the large number of muscles relative to the number of joints requires the controller, our central nervous system, to choose how each muscle contributes to the forces that result in the intended movement. For healthy people, it seems reasonable to assume that we control our muscles following an optimality principle: to minimize the amount of metabolic energy spent on the task. However, a disease, pain or instability are likely to influence a patient’s control strategy; muscle control might be less optimal and more, or less, variable, depending on a person’s ability or need to control force production. Therefore, the general aim of this thesis was to explore the variability in motor control of the musculoskeletal dynamics during walking through a stochastic modelling approach.
Firstly, I discussed the theoretical framework to model human movement dynamics and the current efforts to verify and validate musculoskeletal models, with the aim to quantify the errors in their predictions. Secondly, I aimed to explore the influence of motor control on the mechanical load experienced by the joints of the lower limb during level walking. An optimization approach to motor control showed that alternative motor control strategies have the potential to reduce the loading in the knee and the hip, but not in the ankle, during level walking. These results suggest that neuromuscular rehabilitation can be targeted as a conservative treatment when the mechanical load on joints is a determinant of the onset and/or progression of a disease. However, these alternative motor control strategies come at a cost of a moderate increase in the loading at non-targeted joints. Subsequently, the assumption of a lightly sub-optimal motor control strategy to predict knee contact forces, through a stochastic approach to model motor control, captured the measured intra-subject variability in these forces during multiple gait cycles of a patient with a knee replacement. Therefore, the assumption of sub-optimal control can predict a range of plausible joint contact forces, representative of the uncertainty in terms of measurement inaccuracies, modelling errors and inherent variability, which is likely to contain the true force. However, if a higher accuracy of predicted muscle and joint contact forces is required or in case of severely sub-optimal motor control, I believe the only solution is to include an explicit model of motor control. A refined mechanistic model would allow for the differentiation between hierarchical levels of motor control, as proposed by Bernstein, such as the involuntary spinal control and the cognition-driven anticipatory control.
