A Closed-Loop Method to Identify EMG-Torque Dynamics in Human Balance Control.

Abstract

Human balance control requires continuous modulation of ankle torque by central and spinal activation of the ankle muscles combined with the intrinsic mechanical stiffness of the joint. These components appear together and cannot be measured separately. This work presents a novel multiple-input, single-output, closed-loop identification method that decomposes the ankle torque in human balance control into its central, stretch reflex, and intrinsic components. The method models separate transfer functions for each EMG-torque relation for central and stretch reflex mechanisms and estimates the ankle intrinsic torque from the residuals. The method uses only EMG measurements, requires no kinematic data, and has few parameters, resulting in robust performance. Application of the method to perturbed standing data from two healthy subjects demonstrated that the central and stretch reflex torques accounted for 80-93% of the ankle torque variation, while the intrinsic stiffness was responsible for most of the remaining torque.