Device-embedded accelerometry complements neural signals for tracking parkinsonian motor states

Area of Science:

• Neuroscience
• Biomedical Engineering
• Neurology

Background:

• Adaptive deep brain stimulation (aDBS) uses physiological biomarkers to adjust therapy for Parkinson's disease (PD).
• Neural biomarkers may be affected by stimulation, impacting closed-loop control accuracy.
• Current methods for tracking PD motor state during aDBS require improvement.

Purpose of the Study:

• To evaluate the efficacy of device-embedded accelerometers in tracking Parkinsonian motor state during continuous subthalamic nucleus (STN) deep brain stimulation (DBS).
• To compare the performance of accelerometry-derived features against neural biomarkers for decoding motor symptoms under stimulation.
• To identify robust biomarkers for next-generation adaptive DBS systems.

Main Methods:

• Analysis of over 1,900 hours of chronic recordings including STN neural activity, cortical activity, and device-embedded accelerometry.
• Simultaneous recording of continuous wearable assessments for bradykinesia and dyskinesia.
• Comparison of motor symptom decoding accuracy using accelerometry-derived features versus neural features across different stimulation conditions.

Main Results:

• Accelerometry-derived features robustly tracked motor symptom severity (bradykinesia and dyskinesia) across all stimulation conditions.
• Accelerometry outperformed neural features, particularly STN beta power, in decoding motor state.
• STN beta power's utility was limited by conflating periodic and aperiodic neural processes, with periodic activity showing reduced coupling to symptoms under stimulation.

Conclusions:

• Device-embedded accelerometry provides a robust and stable behavioral biomarker for tracking Parkinson's motor state during deep brain stimulation.
• Neural biomarkers, like STN beta power, exhibit differential robustness and may be less reliable under active stimulation.
• Accelerometry is a promising candidate biomarker for enhancing the performance and reliability of future adaptive DBS systems.