Computational methods for improving estimates of motor unit twitch contraction properties.

Estimates of mechanical properties of human motor units have usually been made indirectly, using the technique of "spike-triggered averaging" (STA). In this method, a single motor unit action potential is used to synchronize the accumulation of an ensemble average of correlated force transients. However, under most realizable conditions, these transients are recorded during periods of sustained motor unit discharge, in which each motor unit is producing a partially fused tetanus. Therefore, the STA technique extracts the characteristics of the unfused force transient, instead of the desired single motor unit twitch. Although the STA method has been widely used, there is as yet no well-established relation between the force transient in the unfused tetanus, and the twitch contraction properties of the motor unit. To evaluate the accuracy of the STA as a measure of the motor unit mechanical properties, we applied two types of muscle models to the force transients recorded in an unfused tetanus, using data derived from experiments in which the response to a single twitch was also recorded. Our objective was to see whether accurate predictions of single motor unit mechanical characteristics are possible, working backward from the STA. The models chosen for this task were a linear second order model, and the distribution-moment (DM) model. These model predictions were then compared with the STA response, and with the twitch properties of the individual motor units. We also evaluated the utility of extrapolating the initial slope of the STA backward to improve the accuracy of the mechanical estimates. The results of our simulation suggest that there is no straightforward relation between the characteristics of the unfused tetanus and the mechanical properties of the single twitch. Although our attempts to predict the properties of the single twitch from the STA were only partly successful, the results of the simulations were far more accurate than those derived from the STA alone. Because the errors in the use of the STA method were so substantial, we would urge that the STA technique be used with great caution as a measure of twitch contraction properties, unless accompanied by appropriate simulations of muscle mechanical behavior.