Optimization reduces knee-joint forces during walking and squatting: Validating the inverse dynamics approach for full body movements on instrumented knee prostheses

Due to the redundancy of our motor system, movements can be performed in many ways. While multiple motor control strategies can all lead to the desired behavior, they result in different joint and muscle forces. This creates opportunities to explore this redundancy, e.g., for pain avoidance or reducing the risk of further injury. To assess the effect of different motor control optimization strategies, a direct measurement of muscle and joint forces is desirable, but problematic for medical and ethical reasons. Computational modeling might provide a solution by calculating approximations of these forces. In this study, we used a full-body computational musculoskeletal model to (1) predict forces measured in knee prostheses during walking and squatting and (2) to study the effect of different motor control strategies (i.e., minimizing joint force vs. muscle activation) on the joint load and prediction error. We found that musculoskeletal models can accurately predict knee joint forces with an RMSE of