Subject-specific musculoskeletal models and computational walking models in general have come a long way in improving clinical treatment of walking disorders. But more accurate predictions of such forces and their locations can help in designing knee replacements to recover normal walking.
This project aims to predict muscle forces for treadmill gait trials when varying speeds and analyse the differences in those muscle forces. Three walking trials with different speeds were studied in this project. Two approaches were used to predict forces. In Approach A, knee contact force information was used as input of the algorithm, and in Approach B, these data was used only to validate the results. An OpenSim musculoskeletal model of the right leg was used to obtain inverse kinematics and inverse dynamics data, and muscle length and moment arms.
The algorithm to estimate muscle forces consisted of a two-level nested optimization. The outer level optimizes the time-independent parameters and the inner level optimizes the time-dependent parameters. Kinematics and ground reaction force data used in this project were obtained from the fourth grand challenge competition to predict in vivo knee loads.
Muscle force estimation values obtained in Approach B (usual case) were significantly different from Approach A (unique case) for most muscles. The results from this study reinforce results of previous studies. Medial and lateral force distribution was also analysed. The muscles with the maximum and minimum differences in mean forces for the three different speeds were identified and possible reasons for these differences were discussed. |