SUBJECT-SPECIFIC AND FUNCTION-SPECIFIC APPLICABILITY OF A HAND-FOREARM MUSCULOSKELETAL MODEL

Abstract

There is ambiguity on the quantitative validity of the predictions revealed by musculoskeletal models for muscle forces. This study investigated the consistency between the predictions of a musculoskeletal model of the upper limb and the experimental data for a number of different subjects and functional tasks. Six normal subjects performed isokinetic eccentric or isotonic concentric contraction tests of the wrist muscles in well controlled conditions, using a robotic apparatus, and the net joint torque and angular velocities, as well as the surface electromyograms (EMG) signals of the muscles, were recorded. The experiments were then simulated using a parametric musculoskeletal model of hand and forearm, scaled for each subject. Muscle activation was determined based on the optimization of a polynomial or min-max objective function. Comparison of the model predictions with the normalized EMGs revealed more consistent results for the polynomial objective function. In general, the model could reflect the change of the net joint torque on the activation level, but the effects associated with the velocity of movement were not sufficiently reflected. Stronger correlations between model predictions and normalized EMGs, were obtained for subjects with smaller body mass indices (BMIs), 0.738, and those who experienced higher muscular excitations, 0.908. The sensitivity analysis of the model also indicated a high sensitivity (9.5%) to the joint torque but a minimal sensitivity (0.3%) to the joint velocity. The sensitivity indices for the muscle’s length and physiological cross-sectional area (PCSA) were 11.2 and 5.3%, respectively. It was concluded that our musculoskeletal model was highly sensitive to the anatomical data, as well as the muscles’ excitation levels, but not to the contraction velocity.