Identification of nonlinear model of ankle joint dynamics during electrical stimulation of soleus.

The mechanical impedance of the ankle joint was estimated in two conditions: during submaximal surface electrical stimulation of the soleus muscle, and with no stimulation applied. Both neurologically intact (n = 5) and spinal cord injured subjects (n = 4) were used. The mechanical impedance was measured by applying angular step and constant velocity (13-100 degrees s-1) perturbations at 10 degrees to the ankle and measuring the resulting changes in torque. A five-element lumped model consisting of an inertial element, a parallel elastic element, and an elastic element in series with a viscous element and a pure tension generator produced a good fit for predicting the compliance characteristics of the ankle for both the relaxed and stimulated conditions. The elastic elements were piecewise linear with different values for the dorsiflexion and plantarflexion directions. The viscous element was velocity-dependent and it decreased in value as the velocity increased. The average torque error between the measured and model's response during soleus stimulation was 10.56% for the dorsiflexed and 11.93% for the plantarflexed perturbations. However, the average error was skewed by several subjects who had excessive error, due to volitional intervention or flexor withdrawal reflex. The average model error for the perturbations without stimulation was 7.12% for dorsiflexed and 5.58% for plantarflexed perturbations.