Objective: To explore a control theory modeling framework to classify which Parkinson’s disease patients may benefit from adaptive deep brain stimulation.

Background: Deep brain stimulation (DBS) is a common treatment for Parkinson’s disease, in which tonic stimulation is delivered to basal ganglia nuclei with manually-programmed, fixed parameters. Stimulation remains continuous despite fluctuations in symptoms, which can result in suboptimal stimulation or adverse side effects from excessive stimulation. Adaptive DBS dynamically controls DBS output in real time using a feedback signal that reflects patient clinical state. However, not all patients may be optimally controlled with a given control algorithm design for adaptive DBS. Furthermore, adaptive stimulation programming is labor-intensive and involves testing a large parameter space. Therefore, even successful therapies are unlikely to achieve widespread adoption without modeling approaches that constrain the search space. Here we sought to apply a modeling framework based on control theory to classify whether subjects are amenable to control with adaptive DBS. Using this framework, we aim to 1) identify patients with neurophysiological responses to stimulation that are conducive to adaptive control, and 2) search the DBS parameter space for optimal control settings.

Method: In 4 patients implanted with bidirectional neural interfaces, we generated patient-specific models to classify controllability. For each patient, we analyzed 3-7 hours of local field potentials during adaptive DBS, recorded from permanently implanted quadripolar leads in the subthalamic nucleus. We assessed the frequency responses of input stimulation current and output neurophysiological band power, generating individual transfer functions representative of how each patient’s brain responded to stimulation. We modeled the neurostimulator as a proportional-integral controller, which we used to estimate patients’ controllability and to sweep the adaptive DBS parameter space.

Results: We show that control theory modeling may provide a framework to classify suitable candidates for adaptive DBS and streamline adaptive DBS programming.

Conclusion: Control theory modeling may aid in the development of patient-specific, adaptive DBS. We explored an approach to classify whether patients can be controlled with adaptive DBS, and this framework may streamline programming by narrowing the parameter space for optimal controller function.

« Back to MDS Virtual Congress 2021

MDS Abstracts - https://www.mdsabstracts.org/abstract/patient-classification-for-adaptive-deep-brain-stimulation/