EEG-based adaptation of robotic neurorehabilitation therapies to patients’ motivational and attentional needs
Understanding the underlying mechanisms of motor learning – i.e. the acquisition of new motor skills–is crucial to develop novel ways to improve neurorehabilitation therapies. Motor learning is mediated by “extrinsic” practice-related factors, such as task’s goal, training intensity, number of movement repetitions or available feedback during training. However, motor learning processes are also mediated by “intrinsic” subject-related cognitive factors, such as motivation, task understanding and attention.
The aim of this project is to get a better insight into the brain-related intrinsic factors during therapy in order to customize and optimize therapies to each subject’s special needs at each stage of the recovery process.
Electroencephalography (EEG) will be used to identify neurocognitive markers underlying motor learning, namely brain potentials reflecting attentional task engagement and motivation-driven oscillations. Robotic motor learning paradigms will be evaluated based on their effect on these neurocognitive markers, e.g. by modulating the level of robotic support and/or guidance. Finally, using machine learning algorithms we will close the loop: real-time assessment of the neurocognitive markers during motor training will be employed to select and adapt the optimal robotic therapy according to the subject’s needs.
We expect that a more informed (human or robotic) therapist would be able to accelerate and increase recovery after a cerebral-vascular accident.
Keywords: neurofeedback, neurorehabilitation, machine learning, neural correlates, motor learning