People with unilateral transfemoral amputations often demonstrate compensatory movement patterns while using socket-type prostheses, resulting in asymmetrical joint loading that increases the risk of developing secondary comorbidities such as osteoarthritis or low-back pain. Bone-anchored limbs (BALs), which attach prostheses directly to the residual limb using osseointegrated implants, have shown promise in improving mobility, muscle function, and overall quality of life. However, there remains a need to better understand the biomechanical outcomes of BAL use. The objective of this dissertation was to assess the longitudinal biomechanical effects of transfemoral BAL use and utilize predictive modeling tools to aid clinical application.

Three Specific Aims guided this work. The first aim developed subject-specific musculoskeletal models to evaluate longitudinal changes in gait biomechanics before and after BAL implantation. This included a novel six degree-of-freedom residuum-socket interface for socket-type prosthesis users. Specific Aim 1 showed that BALs improved muscle function and forward propulsion strategies, but also highlighted that asymmetric joint loading conditions persisted. In the second aim, an optimal control framework was implemented to generate subject-specific gait patterns, joint loading conditions, and muscle activity using BAL models and experimental data. These results were robustly validated to generate solutions to act as baselines for simulated rehabilitative interventions. Specific Aim 2 included test cases investigating the movement pattern and joint loading changes from simulating varied weakened and strengthened conditions of hip abductor muscles. Additionally, a movement pattern retraining intervention was simulated to determine feasibility of symmetrical movement for transfemoral BAL users, and the biomechanical changes resulting from achieving symmetry. Finally, to address limitations of laboratory-based modeling, the third aim used probabilistic analysis to assess the sensitivity of predicted joint moments to perturbations in physical parameters of Hunt-Crossley contact spheres;​ within an optimal control framework driven by wearable sensor kinematic data.

Collectively, this work presents a range of computational tools (including musculoskeletal modeling, optimal control, and probabilistic analyses) which could potentially aid in promoting clinical accessibility to computational biomechanics to inform high-cost clinical trials and help develop patient-specific rehabilitation strategies for transfemoral BAL users.