This thesis aimed to estimate the effect of demographic and biomechanical preoperative variables on total shoulder arthroplasty (TSA) outcomes. To achieve this, various biomechanical variables were calculated using a clinical database comprising normal and pathological subjects. The process involved segmenting and landmarking bones and muscles, measuring morphology, determining joint and muscle forces, and assessing stress/strain within the bone and bone-implant interface. This study was a collaboration among Lausanne University Hospital, the University of Bern, and EPFL.

For morphological analysis, CT scans were used to predict scapula and humerus surfaces, along with landmarks. Key anatomical measurements, such as glenoid orientation, bone mineral density, acromion positioning, and rotator cuff degeneration, were computed. These measurements were automated for the entire database using Python, with a web application developed for visualization and calculation.

Muscle and joint forces were computed from open-source musculoskeletal models. To enhance efficiency, a deep learning model was trained to predict joint forces based on patient characteristics and activities. This model was approximately 1,000 times faster than traditional musculoskeletal methods, improving the workflow's integration.

An automated finite element model for anatomical and reverse TSA was developed. Using CT-based material properties, the finite element model analyzed stress/strain distributions within the bone. Bone and implant geometries were meshed, and musculoskeletal data defined the force inputs, ensuring an efficient, fully automated process controlled via Python.

Causal inference was used to analyze how demographic and biomechanical factors affected TSA outcomes. Directed acyclic graphs (DAGs) outlined assumptions, while do-calculus identified adjustment sets for analysis. Bayesian statistics quantified the causal effects. This framework, encompassing morphological, musculoskeletal, finite element, and statistical analyses, offers a structured approach to enhance TSA outcomes while integrating preoperative planning.