Automated Daytime Adaptive Deep Brain Stimulation Parameter Optimization in Patients Implanted With Percept Neurostimulator

Résumé succinct

Parkinson's disease (PD) affects \~1% of people over 60 years old, is highly disabling and represents a large economic burden. While dopaminergic medications effectively treat motor symptoms early in the disease, most patients develop complications, including motor fluctuations and dyskinesias, which can be partially managed by deep brain stimulation (DBS). This surgical therapy consists of delivering continuous electrical stimulation through electrodes permanently implanted in basal ganglia nuclei, with a pulse generator and battery unit implanted in the chest. However, conventional DBS therapy is delivered with constant stimulation parameters, referred to as constant deep brain stimulation (cDBS), that are unresponsive to patient activities or to variations in the severity of symptoms during daily life. This leaves many patients under- or over-stimulated during parts of the day. To address the shortcomings of cDBS, adaptive DBS (aDBS) uses real-time detection of neural signals to automatically adjust stimulation amplitude or other parameters in response to patients' dynamic clinical needs. aDBS was approved by the U.S. Food and Drugs Administration (FDA) for clinical treatment of PD in the Percept PC and RC (Medtronic) device in February 2025. Fully leveraging this therapy in the real world is limited by technical challenges, in particular the fact that: while the investigators developed a consistent pipeline for implementing aDBS, there were several critical control parameters that strongly influenced algorithm performance and required prolonged trial-and-error based testing, to achieve successful control. In this new study, the investigators seek to significantly extend this work and address the major barriers to widespread, easy adoption of aDBS by groups without specialized knowledge of neurophysiology or feedback control. Here the investigators aim to test an automated, data-driven pipeline for the recommendation of the adaptive control parameters.