Linear and nonlinear features of EEG microstate associated with insomnia

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomarkers

Background:

• Insomnia is associated with EEG microstate abnormalities, but nonlinear metrics are understudied.
• Previous research primarily used linear features to quantify these changes.

Purpose of the Study:

• To compare linear and nonlinear EEG microstate features in insomnia.
• To assess their utility in machine learning-based insomnia classification.

Main Methods:

• Resting-state EEG data from 55 insomnia patients and 67 healthy controls were analyzed.
• Linear (duration, occurrence, coverage, transition probability) and nonlinear (entropy rate, Lempel-Ziv complexity, Hurst exponents) microstate features were extracted.
• A support vector machine was trained using these features for classification.

Main Results:

• Insomnia patients exhibited altered microstate durations, occurrences, and transition dynamics compared to controls.
• Nonlinear analysis revealed differences in entropy rate and excess entropy between groups.
• Integrated linear and nonlinear features achieved 82.8% accuracy in classifying insomnia.

Conclusions:

• Both linear and nonlinear EEG microstate features reflect aberrant brain network dynamics in insomnia.
• These features show potential as reliable biomarkers for insomnia diagnosis.