Using Quantified Motor Behavior Outcomes to Improve Deep Brain Stimulation in Parkinson’s Disease

Abstract

Deep brain stimulation (DBS) is a highly effective therapeutic option for Parkinson’s disease (PD). However, it can take 50 or more hours to obtain stimulation settings that optimally treat a patient’s symptoms. Additionally, axial symptoms, such as gait, are not adequately treated in the long term. In my work I explore the use of quantified motor behavior outcomes to reduce the time needed to obtain optimal stimulation parameters, and to develop a novel stimulation delivery approaches to better treat gait. First, I tested a Bayesian optimization approach to quickly and accurately model the input/output response of rigidity to stimulation frequency. I found, for PD patients that have a high degree of rigidity, Bayesian optimization models their response needing fewer samples than a traditional trial-and-error approach. Next, I tested a novel closed-loop stimulation delivery approach that delivered short duration pulse trains at specific phases of gait. I found that the patients that respond strongly to this type of stimulation delivery have a worse gait with their clinical settings. Overall, many patients saw small changes to their gait with this approach. Lastly, I analyzed the effects of turning stimulation on and off on gait. I found that a repeated measures of gait with short duration, 1 minute, wash-in and -out can detect significant changes. This is in contrast to previous reports that significant changes are only seen between 30-60 minutes. Through these studies I demonstrate the use of quantified motor behavior outcomes to improve DBS for PD.