Prediction of Muscle Activation Patterns During Postural Control Using a Feedback Control Model
Lockhart, Daniel Bruce
MetadataShow full item record
Neural mechanisms determining temporal muscle activity patterns during postural control are not well understood. We hypothesize that a feedback control mechanism can predict both temporal extensor muscle EMG and CoM kinematics (acceleration, velocity, and displacement) during postural perturbations before and following peripheral neuropathy to group I afferents induced by pyridoxine intoxication. We introduce a feedback model for analyzing temporal EMG patterns that decomposes recorded electromyogram (EMG) signals into the sum of three center of mass (CoM) feedback components. EMG and CoM kinematics during postural responses due to support surface translations were measured before and 14 days after somatosensory loss in cats. We successfully predicted EMG before and after peripheral neuropathy by modeling a standing cat as an inverted pendulum and decomposing temporal EMG patterns using a simulation with time delayed feedback loop of CoM kinematics. This model accounts for over 60% of the total temporal variability of recorded extensor EMG patterns. Feedback gains for acceleration, velocity and position necessary to predict EMGs before and after sensory loss were different. For intact animals, more that 90% of the initial burst of EMG were due to CoM acceleration feedback, while later portions were due entirely to velocity and position feedback. After peripheral neuropathy, the initial burst was absent and the acceleration gain was significantly reduced when compared to the acceleration gain of intact animals for extensor muscles (p lt 0.05). By successfully decomposing EMG into three kinematic gains, a quantitative comparison of temporal EMG patterns before and after peripheral neuropathy is possible. The reduction of acceleration gain in sensory loss cats suggests that group I afferents provide necessary information that is used as acceleration feedback.