Advancing Knowledge of Upper Limb Motor Control Through Electromyography

The rehabilitation of motor functions of subjects affected by neurological disorders and amputation will benefit from any advance in our knowledge of the neuromechanical principles of motor control. Electromyography (EMG) holds the possibility to record the electrical activity originating with muscle contraction, thus offering the opportunity to quantify the mechanical direct consequences of the motor command. This project aims to characterize motor production starting from EMG signals in order to: 1) Quantitatively and objectively assess the impact of upper limb assistive devices on motor production and their potential benefits in the rehabilitation of individuals with neuromotor impairments; 2) Explore EMG-driven models that holds potential for the control of rehabilitative devices. The first objective is addressed through a use case employing the muscle synergy framework to investigate the modular organization of motor control during a standardized reaching task performed both with and without robotic assistance from the Float upper-limb exoskeleton. The analysis is initially conducted on healthy participants to establish baseline synergy structures and to examine the extent to which these patterns are altered by robotic assistance. Subsequently, the same approach is applied to chronic post-stroke individuals to assess whether robotic assistance could induce short-term modulation that could promote the restoration of physiological motor modules. This analysis provides insight into the effects of robotic assistance on the modular organization of anterior reaching movements and highlights promising influences of the Float exoskeleton on motor modularity in post-stroke individuals. Importantly, these findings have the potential to inform clinical practice by introducing objective, physiologically grounded metrics for evaluating the effectiveness of technology-assisted rehabilitation. This represents a meaningful advancement over conventional clinical scales, which are predominantly observer-dependent and are often limited by poor inter- and intra-rater reliability, as well as by floor and ceiling effects. The second objective is addressed through another experimental framework aimed at leveraging EMG decomposition techniques and their capability of reconstructing single motor unit activity from global surface EMG recordings. The resulting motor neuron activity patterns represent promising command signals for the control of rehabilitative devices, as they encode the central motor commands conveyed by innervating alpha-motoneurons. Within this context, the activity of individual motor units from the wrist extensors was tracked during goal-oriented reaching, grasping, and object repositioning tasks involving differently shaped objects. In parallel, the feasibility of a hand-gesture classifier based on motor unit activity was explored. The results obtained from these analyses support the feasibility of leveraging motor neuron activity as a physiologically meaningful control signal. At the same time, they highlight the current technological and methodological limitations that hinder the immediate translation of these approaches into clinical practice. Together, the findings of this PhD work delineate both the potential and the challenges of EMG–based tools for neurorehabilitation, providing a foundation for future investigations aimed at bridging the gap between experimental neuroengineering solutions and clinically deployable rehabilitation technologies.