Functional and Robust Human-Machine Interface for Robotic-Assisted Therapy of the Shoulder after Stroke

The majority of individuals with stroke experience problems with the upper extremity, of which paresis is most common. The use of robotics in upper limb therapy is increasingly popular, as it can deliver intensive and functional arm rehabilitation. This thesis describes the development of a functional and robust myoelectric control interface for the pneumatic shoulder RehaARM robot, using electromyographic biofeedback, and the design of a therapeutic approach for shoulder treatment after stroke with this technology. The therapeutic approach has been evaluated in a phase II clinical trial . The developed interface goes beyond the state-of-the-art technologies for upper limb therapy in clinical research because it is capable of online myoelectric control of the robot with three degrees of freedom (DoF), supporting shoulder movements. Myoelectric control has been applied only in systems providing 1-DoF movements for the elbow or wrist. The developed RehaARM system surpasses previously developed systems in the number of actuated DoF. More importantly, the developed myoelectric control interface is transferable to other active-assistive robots for upper limb therapy. The thesis also describes novel neurophysiological measurements for evaluating the effectiveness of the treatment with the RehaARM system and presents their correlation with the most commonly used clinical impairment and activity upper limb scales. The resulting system has been tested on both healthy volunteers and stroke patients and has been compared with the commercially used torque (force) control. Our experimental studies confirm the benefit of the developed myoelectric control interface for the RehaARM robot. RehaARM with myoelectric control achieved equally desirable effects on muscle activation, namely synergistic activation (muscle recruitment) and modulation of activation levels as torque (force) control. Results showed that severely, moderately and mildly impaired patients were able to successfully control the RehaARM system with myoelectric control with greater ease than torque control for task-specific shoulder training. Myoelectric control was used by severely impaired patients who have very low residual voluntary forces which are rarely detectable by commercially available robots. For all patients, there was a monotonic increase in the movement completion rate over multiple sessions that showed improvement in voluntary control. The therapeutic approach with the RehaARM myoelectric interface significantly improved both motor control (FMA-UE) and activity (FIM) scores. The difference between effects of robotic and conventional therapy alone – comparable dose and dosage – was not significant. This is in agreement with the evidence found in the scientific literature. Based on the FMA-UE scores, the sample sizes needed to determine the effectiveness of the treatment were similar for conventional therapy (n = 27, p<0.05; power = 0.8) and RehaARM therapy (n = 28, p