Objective: To demonstrate the proof-of-concept of a closed-loop deep brain stimulation (DBS) policy designed to reduce slowness in Parkinson’s disease (PD) by increasing stimulation amplitudes during movement.

Background: Subthalamic DBS is an established treatment for alleviating symptoms of PD, including bradykinesia or slowness of movement. Conventional DBS (cDBS) involves constant stimulation amplitudes, which may lead to sub-optimal treatment due to symptom fluctuations. Adaptive DBS (aDBS) represents an emerging paradigm where stimulation levels are modulated based on feedback signals, potentially improving therapeutic outcomes.

Method: We implanted bilateral Medtronic RC+S neurostimulators and electrocorticography arrays in a patient with PD. Recordings were made from the subthalamic nucleus and sensorimotor cortex while forearm speeds were monitored using wrist-worn accelerometers. We developed and validated an aDBS policy that increased stimulation during forearm movement, both in a short, supervised tapping task and over an 18-day period in an unconstrained, naturalistic setting. The study included two control conditions to which the patient was blinded: cDBS and an inverse aDBS policy where stimulation levels were lowered during movement.

Results: We constructed a machine learning pipeline that identified cortical beta (16-32 Hz) and broadband gamma (52-98 Hz) as the most predictive power bands for movement decoding (balanced accuracy: 0.83 ± 0.03). During the tapping task, the aDBS policy significantly improved forearm speed in both hands compared to the inverse control [Figure 1]. For the left hand, the aDBS policy also produced higher tapping speeds than cDBS. Over the 18-day period, the aDBS policy significantly reduced bradykinesia compared to cDBS, as measured by Parkinson’s KinetiGraph (PKG) watches [Figure 2] and self-reports [Figure 3]. There was no difference in tremor severity between the policies. PKG watches noted a slight increase in dyskinesia severity with aDBS, which was not discernible to the patient based on self-reports. Overall, the patient reported a significant preference for the aDBS policy.

Conclusion: Intracranial neural signals enabled the accurate prediction of real-time, naturalistic forearm movement. Using this as a control signal for closed-loop DBS therapy led to significant improvements in motor function compared to conventional DBS in both acute and chronic settings.

Figure 1: Tapping speeds for left and right hands.

Figure 2: PKG symptom severity scores.

Figure 3: Patient self-ratings.

References: 1. Dixon, T. C. et al. Movement-responsive deep brain stimulation for Parkinson’s Disease using a remotely optimized neural decoder. Preprint at https://doi.org/10.1101/2024.08.14.24312002 (2024).

2. Darbin, O. et al. Subthalamic nucleus deep brain stimulation driven by primary motor cortex γ2 activity in parkinsonian monkeys. Sci Rep 12, (2022).

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/decoding-naturalistic-movement-for-adaptive-deep-brain-stimulation-in-parkinsons-disease/