Detecting changes in dynamical structures in synchronous neural oscillations using probabilistic inference.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Network Science

Background:

• Cognitive functions and learning are linked to dynamic brain network activity.
• Existing methods for detecting changes in dynamic brain structures are insufficient.
• A need exists for efficient methods to analyze neural data for dynamic network changes.

Purpose of the Study:

• To develop and validate a novel model-based approach for detecting change points in dynamical brain network structures.
• To quantify temporal changes in dynamic brain networks using Bayesian inference and information-theoretic criteria.
• To establish a method for revealing the neural basis of dynamic brain networks.

Main Methods:

• Combined model-based network estimation with a phase-coupled oscillator model.
• Employed sequential Bayesian inference, using prior and posterior distributions to quantify network changes.
• Utilized Kullback-Leibler divergence as an index to measure changes in dynamical network structures.
• Validated the method using numerical simulations and electroencephalography (EEG) data.

Main Results:

• The Kullback-Leibler divergence accurately indicated changes in dynamical network structures.
• The proposed method successfully estimated directed network couplings.
• Change points in dynamical network structures were reliably identified in both simulated and real neural data.
• The method demonstrated efficacy in analyzing electroencephalography data.

Conclusions:

• The developed method provides an efficient way to detect change points in dynamic brain networks.
• This approach offers a robust tool for quantifying temporal variations in neural connectivity.
• The findings suggest the method can elucidate the neural underpinnings of cognitive processes and learning.