Estimating Human Upper Limb Impedance Parameters From a State-of-the-Art Computational Neuromusculoskeletal Model

The human neuro-musculoskeletal system constantly deploys passive (e.g., posture adjustment) and active (e.g., muscle co-contraction) control strategies to regulate upper limb impedance and stability while interacting with the outside world. Upper limb impedance has been assessed through in vivo experiments and model-based simulations. The experiments are practically limited to small samples of able-bodied subjects and few limb postures, and model-based approaches have mostly used simplified upper limb models. Our objective was to develop and validate a computational approach to estimate upper limb impedance parameters - stiffness, viscosity, and inertia - at the endpoint (i.e., hand) using a neuromusculoskeletal model with realistic geometry. We added a planar manipulandum to an existing upper limb model implemented in OpenSim (version 3.3) and used contact modeling to attach the manipulandum's handle to the musculoskeletal model's hand. The hand was placed at several locations lateral to the shoulder joint along anterior/posterior and medial/lateral axes. At each location, during forward dynamics simulations, the manipulandum applied small perturbations to the hand in eight different directions. The spatial variation of the computed, model-based impedance parameters was similar to that of experimentally measured impedance parameters. However, the overall size of the stiffness and viscosity components was larger in the model than from experiments.Clinical Relevance- Computational modeling and simulations can estimate upper limb impedance properties to complement and overcome the limitations of experiments, especially for clinical populations. The computational approach could ultimately inform new interventions and devices to restore limb stability in people with shoulder disabilities.