Towards Personalized Motor-Restoring Technologies: Characterizing Neural Data-Driven Alpha-Motoneuron Model Parameters

In vivo access to the neural motor code driving muscle activation in humans is essential for developing and controlling effective motor-restoring wearable technologies. Partial estimates of the neural motor code can be non-invasively derived from high-density electromyography (HD-EMG) decomposition. These represent the output produced by a subset of alpha-motoneurons (MNs) innervating the recorded muscle. A key step towards accessing the complete MN pool, thus enhancing the neural information derived from HD-EMG, is creating personalized MN models able to capture the neural properties of each individual. However, the model parameters capturing in vivo human MN activity remain unknown. Therefore, this work characterized the sensitivity of MN model parameters to match spike events when driven by HD-EMG-derived partial estimates of the synaptic input. Additionally, we evaluated how optimization of different parameter sets allowed reproducing in vivo MN activity during a motor task. Altogether, these findings contribute to the development of personalized neuromechanical models that enhance partial estimates of motor code for neural controlled neurorehabilitation technologies.