Recent reviews highlighted the limited results of robotic rehabilitation and the low quality of evidences in this field. Despite the worldwide presence of several robotic infrastructures, there is still a lack of knowledge about the capabilities of robotic training effect on the neural control of movement. To fill this gap, a step back to motor neuroscience is needed: the understanding how the brain works in the generation of movements, how it adapts to changes and how it acquires new motor skills is fundamental. This is the rationale behind my PhD project and the contents of this thesis: all the studies included in fact examined changes in motor control due to different destabilizing conditions, ranging from external perturbations, to self-generated disturbances, to pathological conditions. Data on healthy and impaired adults have been collected and quantitative and objective information about kinematics, dynamics, performance and learning were obtained for the investigation of motor control and skill learning. Results on subjects with cervical dystonia show how important assessment is: possibly adequate treatments are missing because the physiological and pathological mechanisms underlying sensorimotor control are not routinely addressed in clinical practice. These results showed how sensory function is crucial for motor control. The relevance of proprioception in motor control and learning is evident also in a second study. This study, performed on healthy subjects, showed that stiffness control is associated with worse robustness to external perturbations and worse learning, which can be attributed to the lower sensitiveness while moving or co-activating. On the other hand, we found that the combination of higher reliance on proprioception with “disturbance training” is able to lead to a better learning and better robustness. This is in line with recent findings showing that variability may facilitate learning and thus can be exploited for sensorimotor recovery. Based on these results, in a third study, we asked participants to use the more robust and efficient strategy in order to investigate the control policies used to reject disturbances. We found that control is non-linear and we associated this non-linearity with intermittent control. As the name says, intermittent control is characterized by open loop intervals, in which movements are not actively controlled. We exploited the intermittent control paradigm for other two modeling studies. In these studies we have shown how robust is this model, evaluating it in two complex situations, the coordination of two joints for postural balance and the coordination of two different balancing tasks. It is an intriguing issue, to be addressed in future studies, to consider how learning affects intermittency and how this can be exploited to enhance learning or recovery. The approach, that can exploit the results of this thesis, is the computational neurorehabilitation, which mathematically models the mechanisms underlying the rehabilitation process, with the aim of optimizing the individual treatment of patients. Integrating models of sensorimotor control during robotic neurorehabilitation, might lead to robots that are fully adaptable to the level of impairment of the patient and able to change their behavior accordingly to the patient’s intention. This is one of the goals for the development of rehabilitation robotics and in particular of Wristbot, our robot for wrist rehabilitation: combining proper assessment and training protocols, based on motor control paradigms, will maximize robotic rehabilitation effects.
|Titolo della tesi:||Understanding motor control in humans to improve rehabilitation robots|
|Data di discussione:||10-feb-2020|
|Appare nelle tipologie:||Tesi di dottorato|