A Dynamic Mutual Information Measure of Phase Amplitude Coupling

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Phase-amplitude coupling (PAC) links slow and fast neural oscillations, crucial for cognition and implicated in neurological disorders.
• Existing PAC quantification methods often use summary statistics or windowed analyses, failing to capture rapid, dynamic changes at high resolution.

Purpose of the Study:

• To introduce a novel dynamic mutual information measure for quantifying time-varying PAC.
• To enable interpretable PAC estimation at single-sample resolution using state-space modeling.

Main Methods:

• Developed a dynamic mutual information measure for PAC.
• Employed a state-space modeling approach with a Gamma generalized linear model (GLM).
• Incorporated a Gauss-Markov process on regression weights for dynamic estimation.

Main Results:

• Validated the method using synthetic data, demonstrating superior performance in tracking dynamic PAC and distinguishing coupled states.
• Successfully applied the technique to sleep EEG data, identifying PAC during sleep spindles.
• Showcased potential for PAC during sleep spindles as a biomarker for conditions like Alzheimer's disease.

Conclusions:

• The proposed dynamic PAC measure offers a powerful tool for neuroscientific and clinical research.
• This method can reveal transient neural dynamics in disease states and inform neurostimulation protocols.
• Potential applications include real-time brain-computer interfaces and advanced neurostimulation strategies.