UC Irvine

Stroke is the leading cause of long-term disability, and many survivors are left with chronic gait impairments. These can limit societal ambulation and lead to a multitude of health complications, making gait rehabilitation a major priority in stroke recovery. Decades of research have focused on addressing post-stroke gait impairment caused by foot-drop. Still, there have been no breakthroughs, leaving clinicians and patients with limited options that often provide only temporary effects. Therefore, novel therapies, based on neurobiological principles of stroke recovery, should be developed to provide long-lasting improvements. Brain-computer interfaces (BCIs) have been suggested as a method of harnessing the brain's neuroplasticity via the Hebbian mechanism. BCIs can directly access the brain, and when paired with other rehabilitation modalities such as functional electrical stimulation (FES) and robotics, Hebbian-like changes can strengthen connections between neural populations, leading to long-lasting improvements. While BCI-based therapies are being developed, the mechanisms underlying them have been challenging to demonstrate in clinical trials. Clinical assessments such as the Fugl-Meyer, stroke scales, gait velocity, and gait endurance have been developed to assess an individual's stroke-related disability and quantify recovery during the rehabilitation process. However, these assessments do not capture the neurophysiological changes that accompany rehabilitation, making it difficult to determine whether neuroplasticity has indeed occurred. While some efforts have tried to include neurological assessments that may provide insight into the brain-muscle connection, most of these assessments are either static or do not involve volitional movement. In this work, we develop a novel assessment framework that addresses these challenges. We use a computational model of the human motor control system to identify the potential impact of post-stroke injury and targets of rehabilitation strategies. This allows us to understand the potential and expected impact of BCI-based rehabilitation specifically, and test that theory. With this knowledge, we then design an assessment method using noninvasive electrophysiological recording systems that utilize EEG, EMG, and kinematic data during dynamic tasks to provide a personalized, comprehensive understanding of neuromuscular connections. We collect baseline data from 19 able-bodied participants and recruit 10 chronic stroke participants who undergo 4 weeks of BCI-FES or BCI-robotic rehabilitation interventions. Based on these experimental datasets, we identify and model how injury impacts different aspects of movement. We then study how these change after rehabilitation and whether we can estimate neuroplastic changes using our novel assessment. This work establishes a neurophysiological assessment framework that can quantify personalized rehabilitation-induced neuroplastic changes, offering new insights into the mechanisms of BCI-based stroke recovery.