Search

Stroke imposes growing economic costs on the United States each on the order of billions of dollars. Many of the costs associated with stroke result from the long-term disabilities a stroke incident often entails. Researchers have thoroughly explored the use of robotics and virtual reality for treating stroke victims to provide more precision, repeatability, and data capture than conventional methods. However, increasing demand for stroke-related rehabilitation creates an impetus for exploring robotics and virtual reality in rehabilitation for the sake of automation and patient independence; automating some or all of the treatment process may reduce the burden on healthcare professionals, lowering the overall cost of treatment. Furthermore, recent initiatives to commoditize virtual and augmented reality have produced compelling technologies that make the abstract notion of independent and personalized treatment systems more concrete. Concretely, this project explores the problem of formalizing and implementing the rehabilitation process for sensory-motor and visual neglect (a particular post-stroke condition) as an automated system. We first reviewed conventional methods for treating neglect, as well as research on the application of robotics and virtual reality to post-stroke rehabilitation. Next, we developed a set of engineering requirements for the project based on safety, existing treatment methods, data capture, and usability. The solution system is designed as software to run on the BKIN KINARM system; a robotic and virtual reality platform designed for neuroscientific research. We then designed an implemented a set of assessment and treatment exercises based on an analysis of conventional methods, an analysis of the host platforms capabilities, and ongoing consultation with healthcare professionals. In addition to the treatment system, we addressed the problem of patient evaluation and monitoring by developing a browser-based data platform for viewing time-series metrics computed from the KINARM’s sensor data. The metrics are designed based on a review and analysis of conventional methods for evaluating progress in stroke rehabilitation. Preliminary data was obtained over a five week treatment and follow-up case study with a stroke patient. The case study data set provides no bases for evaluating the effectiveness of the system for treating neglect. However, it does provide insights around the feasibility of the experimental system, as well as suggestions for refining the research problem and designing future experiments. The data suggests the potential for countering behavioral biases using virtual effects. It also raises questions about the role of subject motivation in producing good rehabilitation outcomes. Final recommendations include redirecting future research away from the exercise-oriented, iterative model of rehabilitation, toward the implicit approach of using virtual effects to counter behavioral biases and re-train during the patient’s normal daily activates.