Beta Burst Characteristic Predict Instantaneous Hand Movement Dynamics in Parkinson Disease: A Novel Biomarker for Closed-Loop DBS

Introduction: Parkinson disease (PD) is a neurodegenerative disorder characterized by bradykinesia, with evidence of increased beta power in cortical and subcortical structures. Biomarkers for closed-loop modulation must assess electrophysiology and predict behavior to appropriately trigger stimulation. Little research has analyzed instantaneous neural signals with even less generating predictive models for movement kinematics. We hypothesize that beta burst characteristics can predict current hand movements and may therefore be a critical signal for therapeutic advances.

We analyzed the dynamics of beta band burst characteristics and their relationship with kinematics of bradykinesia, with an additional focus on the role of causal connectivity between cortical and subcortical regions in mediating bradykinesia.

Methods: Movement and neural data were recorded from the motor (M1), premotor (PM), and internal Globus Pallidus (GPi) of PD patients (N=21). Beta bursts were defined as analytical envelopes of low (12-20 Hz) and high (21-35 Hz) beta oscillations exceeding the 75th percentile threshold. We studied the relationship between bursting activity and hand movement characteristics using Generalized Linear Model (GLM), Linear Mixed Effect Models, and custom MATLAB codes.

Results: High beta band power, burst duration, and bursting rate in M1, PM, and GPi are all correlated with hand movement, specifically in a 100-millisecond window surrounding the movement (p < 0.05, r > 0.1, AIC = 2193, Pearson Correlation Coefficient). Each of these beta bursting characteristics interact together to further bradykinesia. Interactions between the GPi high beta band power with M1 low beta band power modulate behavior (p < 0.05). These interactions can predict hand behavior with a 37% predictive rate.

Conclusion : Bradykinesia is associated with changes in low and high beta oscillation characteristics, both at the local and network level. These signals dynamically predict hand movement dynamics, serving as precise biomarkers to determine ongoing symptomatology for PD patients that serve as novel targets for closed-loop DBS.