Machine versus physician-based programming of deep brain stimulation in isolated dystonia: A feasibility study

Area of Science:

• Neurology
• Neurosurgery
• Biomedical Engineering

Background:

• Deep brain stimulation (DBS) of the internal globus pallidus (GPi) effectively treats dystonia motor symptoms.
• Optimal GPi-DBS programming is challenging due to delayed symptom control, lack of biomarkers, and no defined optimal stimulation site.
• Complex postoperative management with frequent physician follow-ups presents a barrier for medication-refractory dystonia patients.

Purpose of the Study:

• To prospectively evaluate machine-predicted GPi-DBS programming settings against standard clinical programming in dystonia patients.
• To assess the efficacy and efficiency of an algorithm-driven approach for GPi-DBS programming.

Main Methods:

• Developed an algorithm using an anatomical map of GPi stimulation volumes and clinical outcomes to predict optimal stimulation parameters.
• Reconstructed individual patient models to test thousands of settings in silico.
• Conducted a prospective study comparing machine-predicted settings (C-SURF) with clinically derived settings in 10 dystonia patients.

Main Results:

• C-SURF programming achieved a significantly higher dystonia symptom reduction (74.9% ± 15.3%) compared to clinical programming (66.3% ± 16.3%, p < 0.012).
• Total electrical energy delivered (TEED) was comparable between C-SURF and clinical programming.
• Machine-based programming demonstrated superior clinical outcomes in this cohort.

Conclusions:

• Machine-based programming holds significant clinical potential for optimizing GPi-DBS in dystonia.
• This approach can substantially reduce the programming burden and complexity of postoperative management.
• Algorithm-driven programming may improve accessibility and adoption of DBS for dystonia.