Sensitivity of Shoulder Musculoskeletal Model Predictions to Muscle-Tendon Properties

Objective: While the sensitivity of estimated muscle forces to muscle-tendon properties is well documented for the lower limbs, little is known about the shoulder and upper limbs. The purpose of this study was to assess the sensitivity of estimated shoulder muscle forces and scapulohumeral joint force to muscle-tendon properties. Methods: One healthy male participant performed arm flexions and simulated throwing maneuvers. Kinematics were recorded using intra-cortical pins. Muscle forces were estimated using static optimization with the generic delft shoulder and elbow in OpenSim, and scapulohumeral joint forces were calculated fromthe estimated forces. Then, variations from -25% to + 25% of the nominal values of the tendon slack length, the optimal fiber length, the maximal isometric force, and the pennation angle were applied to the musculoskeletal model to compute affected muscle forces and scapulohumeral joint force. Results: The variations in muscle-tendon properties led to changes up to 9.6 N or 174% in the muscle nominal forces. The more sensitive muscles were those that produced the greatest force: the rotator cuff muscles and the prime movers specific to the task. Among the four muscle-tendon properties, the maximal isometric force and the optimal fiber length had the greatest influence on the muscle force variability. Glenohumeral force was slightly influenced by muscle-tendon properties (< 8%). Conclusion: Generic models (i.e., those without personalization of muscle-tendon properties) can lead to misinterpretations of muscle force. Efforts should focus on the maximal isometric force and the optimal fiber length of the rotator cuff muscles and prime movers.