Hybrid Neuromusculoskeletal Modeling

Abstract

Predicting the actions of muscles controlled by the nervous system allows understanding how movement is generated in humans and animals. Computational neuromusculoskeletal models use surface electromyography (EMG) to “drive” simulated muscles to predict the resulting joint moments produced during dynamic tasks. We present a hybrid EMG-driven/EMG-assisted musculoskeletal model of the human lower extremity that addresses the main limitations of current EMG-driven methods. The new model estimates activation patterns for muscles from which EMGs cannot be measured and adjusts experimental EMG recording that may be subject to measurement errors, while calculating joint moments from multiple joints with multiple degrees of freedom. The model ensures dynamically consistent solutions, estimating the dynamics of 34 muscles and the resulting joint moments simultaneously produced around six degrees of freedom in the human lower limb. We describe the theoretical aspects of the proposed methodology and present experimental results that demonstrate the benefits of the new method.

This work was supported in part by the National Institute of Health in the USA under Grant R01EB009351-01A2, the National Health and Medical Research Council in Australia under Grants 628850 and 334151, the Western Australia Medical and Health Research Infrastructure Council, the ERC Advanced Grant DEMOVE.