Analysis of A-phase transitions during the cyclic alternating pattern under normal sleep

Area of Science:

• Neuroscience
• Sleep Science
• Signal Processing

Background:

• Cyclic alternating pattern (CAP) is a non-REM sleep phenomenon characterized by A-phases and B-phases.
• CAP reflects sleep instability and is crucial for understanding sleep processes.
• A-phases (A1, A2, A3) are distinct EEG events separated by basal B-phases.

Purpose of the Study:

• To analyze EEG signal properties during transitions between B-phases and A-phases.
• To investigate complexity, statistical, and frequency characteristics of EEG during CAP.
• To develop a classifier for distinguishing A-phase onsets/offsets from B-phases.

Main Methods:

• Analysis of electroencephalography (EEG) signals from ten healthy sleepers.
• Calculation of standard deviation, energy, sample entropy, Tsallis entropy, and frequency band indices.
• Implementation of a k-nearest neighbors (kNN) classifier for phase discrimination.

Main Results:

• Significant differences (p < 0.05) in EEG signal properties (energy, entropy, frequency indices) between A-phases and B-phases.
• A-phase onset exhibited three times higher energy than B-phase across all sleep stages.
• High classification performance (>80% accuracy, specificity; >70% sensitivity) for distinguishing A-phases from B-phases, except for A3-phase.

Conclusions:

• EEG signal analysis reveals distinct characteristics during CAP A-phases and B-phases.
• Energy in different frequency bands is a key feature for automatic separation of A-phases and B-phases.
• A-phases represent superimposed neural coordination during sleep oscillations.