University of Minnesota
This project aims to establish the feasibility of Bayesian optimization for tuning deep brain stimulation (DBS) to treat gait symptoms in Parkinson's disease (PD) patients. Our primary question is: Can Bayesian optimization of DBS achieve reproducible results within a feasible number of gait measurements? PD patients will be enrolled who have DBS of the subthalamic nucleus (STN) or globus pallidus (GP) in whom at least 3 months have passed since activation of their neurostimulators, for stabilization of clinical stimulator settings. We will apply Bayesian optimization to derive DBS settings which maximally lengthen step length relative to the OFF DBS state.
Patients will undergo on-label DBS for the treatment of Parkinson's disease, at stimulator settings different from their usual settings, but within the FDA-approved range for this indication, based on the output of the Bayesian optimization algorithm. Test settings will be evaluated by a brief period of walking on a treadmill which captures measurements of step length. The resulting step length will be provided to the algorithm to inform the next recommended test setting. This will be repeated approximately 30 times to reach putatively optimal DBS settings.
Inclusion Criteria: * Diagnosis of Parkinson's Disease * DBS in STN or GP (bilateral or unilateral) * At least 3 months after lead implantation Exclusion Criteria: * Inability to walk in the off-med, off-stimulation condition (even with safety harness) * Gait impaired significantly by a condition other than PD, as determined by the PI * Breaks or shorts in active contacts * IPG battery nearing end of life (in patients with primary-cell IPGs) * Females who are nursing or pregnant * Diminished capacity to consent (concluded via UBACC)
DBS within FDA-approved limits and labeling for symptoms of PD
Arctuva estimate
Verify or correct the compensation shown for this listing.