Impact of biofeedback and task-specific training with a robotic hand orthosis on voluntary muscle modulation for rehabilitation post-stroke

PROJECT SUMMARY Robotic devices for hand rehabilitation show promise in improving access to motor training and encouraging functional use of the impaired limb. These devices can provide assistance for daily activities and augment traditional rehabilitation methods. Wearable exoskeletons are a particularly exciting area of research because they could provide therapy beyond the confines of a clinic or laboratory. Our use of surface electromyography (EMG) sensors and intent detection algorithms has enabled individuals with post-stroke hemiparesis to intuitively control a wearable robotic hand orthosis. However, a major barrier for adoption of this or similar devices is excessive spasticity, which is amplified by users’ recruitment of all available muscles when exerting effort to control the robot. This excessive coactivation of muscles when attempting movement patterns is a common complication for stroke. Our work aims to address this problem by using EMG biofeedback, the display of real- time information about muscle activation to the user, to co-train human-robot systems to generate motor patterns when grasping that minimize excessive coactivation. Inspired by studies on visual biofeedback of muscle activity, which have revealed promising results in rehabilitative training, our preliminary work with chronic stroke subjects has indicated that some individuals retain some capacity to change muscle activation patterns in response to EMG biofeedback. The goal of this research is to determine whether EMG biofeedback can be harnessed to help train stroke survivors to modulate muscle activation and generate desired movement patterns with robot assistance while minimizing unwanted coactivation and spasticity. This goal will be accomplished by pursuing two aims. Aim 1 takes an assistive approach to biofeedback and robotic training. We will determine the extent of flexor/extensor decoupling that is achievable when stroke survivors use EMG biofeedback with robotic assistance. We expect EMG biofeedback to aid discrimination and generation of motor patterns that result in the least abnormal coactivation. In Aim 1, subjects will participate in a single-session experiments that reinforce robot-assisted hand movements in alignment with coordinated flexor/extensor activation. Aim 2 takes a rehabilitative approach, and will investigate whether multi-session practice with EMG biofeedback and robotic training produces rehabilitative effects and functional outcomes that persist after the orthosis is removed. To achieve this, we will conduct a multi-session training regimen in which the orthosis requires a progressively higher-fidelity activation signal in order to assist movement completion. The proposed project will provide insights into the progression of human-robot fluency during training and greater understanding of motor learning after stroke. This training complements my development plan by providing an opportunity to work with an interdisciplinary mentoring and collaboration team to pursue a project at the intersection of robotics engineering and stroke rehabilitation, and will ultimately prepare me to lead a translational research laboratory developing patient-centered approaches to build devices to improve motor control in people with neurological disorders.