How well does a musculoskeletal model predict GH-joint contact forces? Comparison with in-vivo data

The Delft Shoulder and Elbow Model (DSEM), a large-scale musculoskeletal model, allows for estimation of individual muscle and joint reaction forces in the shoulder and elbow complex. Although the model has been qualitatively verified previously using EMG signals, quantitative validation has not yet been feasible. In this paper we report on the validation of the DSEM by comparing the GH-joint contact forces estimated by the DSEM with the in-vivo forces measured by a recently developed instrumented shoulder endoprosthesis, capable of measuring the glenohumeral (GH) joint contact forces in-vivo [1]. To validate the model, two patients with instrumented shoulder hemi-arthroplasty were measured. The measurement process included the collection of motion data as well as in-vivo joint reaction forces. Segment and joint angles were used as the model inputs to estimate the GH-joint contact forces. The estimated and recorded GH-joint contact forces for Range of Motion (RoM) and force tasks were compared based on the magnitude of the resultant forces. The results show that the estimated force follows the measured force for abduction and anteflexion motions up to 80 and 50 degrees arm elevations, respectively, while they show different behaviors for angles above 90 degrees (decrease is estimated but increase is measured). The DSEM underestimates the peak force for RoM (up to 38% for abduction motion and 64% for anteflexion motion), while overestimates the peak forces (up to 90%) for most directions of performing the force tasks.