Simultaneous estimation of intrinsic and reflex contributions to muscle dynamics: a modeling study

The purpose of this study is to evaluate three approaches for quantifying the in vivo stretch reflex response to stochastic perturbations. Two recently published nonlinear system identification algorithms for simultaneously estimating intrinsic and reflex dynamics and a linear estimation procedure that quantifies the combined intrinsic and reflex response were compared. The nonlinear algorithms differed primarily in that one was parametric and one was nonparametric. The efficacy of these algorithms was evaluated using simulated data for the human ankle, the human elbow, and the cat soleus. It was found that the nonlinear parametric algorithm separated intrinsic and reflex dynamics only when the model structure assumed a priori matched that of the system under study. The nonlinear nonparametric algorithm adequately identified intrinsic and reflex responses provided that the reflex delay was longer than the decay time for the intrinsic dynamics. This assumption was not valid for the cat soleus model. When the intrinsic and reflex dynamics could not be separated using the reflex delay, the simple lumped model of intrinsic and reflex dynamics provided as much information about the reflex contributions to the muscle response as did the more complicated nonlinear algorithms