A dynamic generator model for motor-related biosignals

Investigation of the human motor system frequently requires the precise determination of motor events indicating transitions between different motor processes currently activated. Frequently, off-line techniques are used for automatic detection of these events from the motor-related biosignals measured in experiments. However, little is known about the accuracy of these methods. This paper describes a new dynamic generator model for motor-related biosignals that fuses knowledge from both biomedical and technical disciplines in order to provide a solid basis for a quantitative assessment of the performance of detection methods. The model comprises higher level control mechanisms of voluntary movement as well as characteristic details of muscle function like its complex nonlinear feedback mechanisms. Further, important morphological and physiological aspects of signal generation and transmission that are peculiar for these signals are considered. The generator is equally suited for modelling surface electromyographic (EMG) signals as well as kinetic signals (e.g. joint angle, muscle force) associated with a particular movement. The performance of the model is demonstrated for the special case of superimposed discrete and rhythmical single-joint movements