Sensor-based control strategies for integration of epidural electrical stimulation in rehabilitation after spinal cord injury

Spinal Cord Injury (SCI) results in damage to neural circuitry connecting the brain to the periphery. Consequently, sensory and motor function is lost, to varying degree, depending on the site and severity of the lesion. More than half of spinal cord injuries result in paraplegia, a partial or complete paralysis of the lower limbs. Despite the emergence of therapeutic interventions in recent decades, outcomes have been incremental, and few have transformed SCI rehabilitation.Epidural electrical stimulation (EES) stimulates paralyzed muscles. Anatomically intact neural circuits below the lesion can be reactivated to a functional state by stimulating afferent fibers in the dorsal roots. Recently, EES enabled independent stepping and improved functional recovery during activity-based rehabilitation. Neuroplasticity, the formation of new functional pathways based on neural activity above and below the lesion, allows for motor control recovery. However, it took months of individualized rehabilitation training to see functional recovery, making widespread application difficult.For the first time in humans, we use biomimetic EES to deliver temporal sequences of spatially selective stimulation trains to the spinal cord. During rehabilitation exercises, we target specific muscle groups to mimic a natural afferent firing pattern. We improve simultaneous descending and ascending activity by using closed-loop motor-intent sensors. We propose three goals to drive clinical translation of biomimetic EES.We first enable EES for locomotor training and outside of a dedicated research environment. Following a systematic review of gait-detection algorithms, we developed a novel neurostimulation platform the functions in closed-loop using inertial measurement units. We further designed a platform intended for independent use of stimulation by the patient. We investigated clinical outcomes of biomimetic EES applied during locomotor rehabilitation and found that performance improved with stimulation and after a few months, participants regained voluntary control over previously paralysed muscles without stimulation.Then we extend EES to patients with trunk impairments and complete leg motor loss. We modified our closed-loop paradigms for people with motor complete SCI and created an electrode lead that covers the thoracic spine. We evaluated biomimetic EES in complete SCI patients during locomotor and trunk rehabilitation. Three participants could walk and control trunk movements after neurorehabilitation.Finally, we incorporate EES into rehabilitation activities across the continuum-of-care. We created a graphical user interface for clinicians to easily create activity-specific simulation programs and synchronized stimulation with common robotic devices. Then we evaluated EES's clinical benefit in conventional and robotic rehab. Individuals could immediately stand, walk, cycle, swim, and move their trunks. Using EES and rehabilitation robotics restored physiologically relevant muscle activity and allowed for more challenging training. Recreational rehabilitation activities, that were otherwise inaccessible, could be performed.Biomimetic EES optimizes conditions for neuroplasticity and shows promise of delivering clinically relevant outcomes in a short timeframe. We thus bring the intervention closer to a clinical reality, offering a new hope to the SCI community.