A Permutation Entropy Method for Sleep Disorder Screening.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Data Science

Background:

• Generalized sleep pathologies exhibit altered brain dynamics.
• Single-channel electroencephalographic (EEG) signals contain potential biomarkers for sleep disorders.
• Current detection methods may require complex setups or sleep-stage annotations.

Purpose of the Study:

• To introduce a novel fractal analysis approach for detecting generalized sleep pathologies.
• To validate the fractal scaling exponent of permutation entropy as a robust biomarker for pathological sleep.
• To assess the method's accuracy and generalizability across different sleep disorders and recording parameters.

Main Methods:

• Utilized two public datasets (Sleep-EDF and CAP Sleep Database) with 200 subjects.
• Computed the fractal scaling of permutation entropy from single-channel EEG signals.
• Analyzed differences in scaling exponents between healthy controls and patients with sleep pathologies.

Main Results:

• Significantly reduced scaling exponents were observed in pathological sleep (mean = 1.24) compared to healthy controls (mean = 1.06, p<0.001).
• Achieved 90% classification accuracy for REM sleep behavior disorder (F1-score: 0.89).
• Maintained 74% accuracy when aggregating various sleep pathologies.

Conclusions:

• The fractal scaling exponent of permutation entropy is a robust, non-specific biomarker for sleep pathology detection.
• The method is compatible with single-channel EEG, independent of sleep-stage annotations, and generalizable.
• This computationally efficient approach could enhance sleep disorder screening and enable earlier intervention, including for neurodegenerative conditions.