Multiscale musculoskeletal modeling of the lower limb to perform personalized simulations of movement

Computational modeling has been used for many decades to inform design and decision-making in several fields of engineering, such as aerospace, automotive, petroleum, and others. However, it still struggles to have a similar impact in fields of medicine, such as orthopaedics. Three of the challenges that have limited the use of computational modeling in the clinical practice and in product development are model validation, personalization, and realism. Validation is a challenge because several internal parameters of the human body, such as muscle forces, are not safely measurable in vivo and, consequently, a thorough comparison between model outputs and experimental measurements is not always possible. Personalization is an additional issue because the inherent variability across a population needs to be accounted for in a model. Finally, the computational burden of simulations performed with a musculoskeletal model limits its level of realism. The purpose of the work presented in this dissertation is to investigate the applicability of state-of-the-art tools, and propose novel approaches to foster an evolution of computational modeling in orthopaedics. Specifically, (1) the reliability of the knee contact force predictions of a musculoskeletal model commonly used in the literature was analyzed using a global probabilistic analysis for three subjects with instrumented implants; (2) subject-specific and activity-specific moment arms of the muscles spanning the knee