Objective: To develop a standardized scoring system design to systematically assess and monitor adaptive algorithm performance over time in Parkinson’s disease (PD) patients.
Background: Deep brain stimulation (DBS) is an established treatment for advanced Parkinson’s Disease (PD) and adaptive DBS (aDBS) offers additional benefits by using physiological or behavioral biomarkers to modulate stimulation dynamically. However, there are currently no automated data-driven methods to quantify the effectiveness of these adaptive algorithms. We propose a novel scoring system for algorithm performance using recorded neural data and wearable scores, enabling structured evaluation of therapeutic impact.
Method: From a PD patient implanted with chronic electrocorticography, subthalamic electrodes and an investigational bidirectional neurostimulator undergoing aDBS for 20 days, we recorded aDBS neural input biomarkers, stimulation parameters and monitored motor symptoms using Parkinson’s KinetiGraph (PKG) watches. We aligned the delivered adaptive stimulation data and the PKG bradykinesia and dyskinesia scores for each day, and computed Spearman correlations. We then evaluated the relationship between daily Spearman correlation coefficients, or rho values, and corresponding patient self-reported daily ratings of algorithm performance and motor symptom severity using linear regression.
Results: We found that 85% (35 out of 40) of the computed Spearman correlations across days (two hemispheres) between delivered stimulation and the PKG scores were significant. The p-values for the Spearman correlations helped us determine which rho values to include for the subsequent regression analysis. Next, we found a significant positive correlation (p=0.0149) between the valid subset of Spearman rho values and patient self-reported daily bradykinesia duration, highlighting their association with a clinically-relevant metric. We then developed an inversion-based scoring method, where lower daily rho value suggests greater algorithmic efficiency.
Conclusion: We propose a novel automated scoring method using the correlation strength between delivered stimulation amplitude and PKG symptom scores. The proposed scoring framework offers a practical method to quantify algorithm effectiveness and can support more standardized evaluation of aDBS therapies.
To cite this abstract in AMA style:
K. Balagula, J. Yao, M. Shcherbakova, A. Hahn, P. Starr, S. Little. Adaptive Deep Brain Stimulation Algorithmic Performance Evaluation Through Symptom Driven Metrics [abstract]. Mov Disord. 2025; 40 (suppl 1). https://www.mdsabstracts.org/abstract/adaptive-deep-brain-stimulation-algorithmic-performance-evaluation-through-symptom-driven-metrics/. Accessed November 20, 2025.« Back to 2025 International Congress
MDS Abstracts - https://www.mdsabstracts.org/abstract/adaptive-deep-brain-stimulation-algorithmic-performance-evaluation-through-symptom-driven-metrics/