Investigating glenohumeral joint forces using a population-based musculoskeletal model combined with a shoulder...

Shoulder anatomy varies between individuals, affecting glenohumeral joint (GHJ) biomechanics. Musculoskeletal (MSK) models can simulate these internal loadings, which depend on anatomical and dynamic constraints. While MSK models are often personalized, population coverage is usually limited. Statistical shape models (SSMs) can capture population-level anatomical variability and investigate its effect on joint forces. However, most SSM-based MSK applications focus on pathological cases and omit one or more shoulder bones. This project developed a population-based MSK model incorporating a three-shoulder-bone SSM (humerus, scapula, clavicle) to assess GHJ force variability and sensitivity to body metrics (height, BMI) and external loads. 3D shoulder bones and height from 61 subjects (29 M, 32F, 36.1 ± 14.3 years) were included in the SSM. The SSM generated 500 instances to non-uniformly scale a generic MSK model. GHJ forces (mean ± SD, extremes) were determined during 90°-abduction and -flexion, with and without a hand weight (0.1-3 kg). Peak forces were regressed against body metrics and hand weight. Population forces matched literature values (90°-flexion: 83.7 ± 17.7 %BW, 90°-abduction: 52.8 ± 9.6 %BW), with forces increasing under load (up to threefold), while extreme cases were linked to body metrics. Unweighted peak forces moderately correlated with body metrics (R² ∼ 0.50), while weighted forces strongly correlated with body metrics and external load (R² ∼ 0.89). Population-based SSM-derived models provide exhaustive statistics and robust regressions through dense sampling of the anatomical parameter space. By investigating larger cohorts, population models extend result generalizability while alleviating clinical data limitations, making them promising tools for in-silico clinical trials.