PROJECT SUMMARY As Parkinson’s disease (PD) progresses, patients develop slowness of movement (bradykinesia), tremor and rigidity. The first-line treatment for PD is the dopamine precursor levodopa. However, long-term use can lead to excessive movement (dyskinesia). To address levodopa-induced dyskinesia, medication dose is adjusted, and deep brain stimulation (DBS) of the basal ganglia is utilized. Without sufficiently lowering the medication dose, DBS can exacerbate dyskinesia. However, the mechanisms with which DBS causes dyskinesia, and how they are modulated by the dopaminergic cycle, are not fully understood. Excessive gamma band (65-90Hz) synchronization in the motor network correlates with dyskinesia in preliminary human studies. This signal has the potential to predict the onset of a dyskinetic episode when identified across a complete set of spectral features with continuous measures of motor fluctuations. During therapeutic DBS of either the subthalamic nucleus (STN) or globus pallidus (GP), the gamma oscillation is often entrained (amplified and modulated) to one-half the stimulation frequency. However, the clinical relevance of entrainment, and how it may interact with across stimulation settings, has not yet been investigated. We predict that stimulation-induced 1:2 entrainment of cortical oscillations (oscillations detected at one-half the stimulation frequency) will therapeutically increase movement, while simultaneously decreasing excessive/pathological movement, otherwise known as dyskinesia. The goals of this study are to use multisite, chronically implanted neurostimulators and continuous wrist-wearable sensors to identify correlates of dyskinesia across a complete set of spectral features and to assess the possible clinical benefit, as it relates to dyskinesia scores measured by the wearable sensors, of stimulation-induced entrainment of these spectral features. First, we will characterize correlates of dyskinesia using hundreds of hours of chronic neural recordings collected in a naturalistic environment across typical dopaminergic cycles with objective dyskinesia monitoring using wearables prior to the initialization of stimulation. The majority of these patients experience dyskinetic motor signs while taking their clinically recommended medications. Then, we will determine the most predictive features of dyskinesia using machine learning models to predict the continuous dyskinesia scores derived from validated wearable algorithms of the sensors worn on the wrist. We hypothesize that periodic gamma components will be most predictive of dyskinetic episodes prior to the onset of stimulation. Next, we will determine the relationship between gamma entrainment, dyskinesia severity, and deep brain stimulation amplitude. To do this, we will record neural data in a similar setting as before. However, during this process, patients will experience deep brain stimulation at the clinically recommended stimulation frequency, and patient...