A realistic upper-limb musculoskeletal model for overall muscle activation estimation

Abstract

The human upper-limb can perform complex tasks thanks to the ability of the central nervous system (CNS) to control and modulate the activation of more than 40 muscles. A deeper understanding of the strategies implemented by the CNS to perform these movements could help develop more effective rehabilitation protocols. Unfortunately, given the large number of muscles acting on the shoulder and the arm, and their positions with respect to each other, a simultaneous electromyographic (EMG) recording of all the upper-limb's muscles is not feasible in practice. Numerical musculoskeletal models could represent a useful alternative approach to gather this kind of information. Here, we develop a new realistic upper-limb biomechanical model which uses a combination of few recorded EMG data (up to max N=16) and an inverse dynamics model to predict the overall upper-limb muscle activations (N=42). With data from five healthy subjects performing 3D upper-limb movements, the predictions from this EMG-assisted model were compared to: 1) a full set of 16 arm muscles EMG data, 2) a model using none of the these recorded EMG data (standard static-optimization). While predicted signals from the EMG-assisted and static-optimization led to a 3.4% and 2.3% error respectively in the resulting moment of joint when fed into the movement dynamics, the EMG-assisted method presented a more physiological sharing of the muscle activations and variations of errors with respect of the activations from EMG signals for non used muscles in a comparison leave-one-out method. These results demonstrate the ability of the proposed musculoskeletal model to be used as a tool to evaluate healthy motion data, and open up to the application of such a strategy to analyze impaired individuals motion data and to compare inter-subject activation behaviours in the future.