Efficient sleep classification based on entropy features and a support vector machine classifier.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Sleep quality is crucial for physical and mental health.
• Efficient sleep stage scoring offers significant healthcare advantages.
• Current methods may lack simplicity or efficiency.

Purpose of the Study:

• To propose a simple and efficient sleep classification method named SC-En&SVM.
• To utilize entropy features and a support vector machine (SVM) classifier for sleep scoring.
• To evaluate the effectiveness and reliability of the proposed method.

Main Methods:

• Entropy features (fuzzy measure entropy, fuzzy entropy, sample entropy) extracted from EEG and EOG signals.
• Application of a multi-class support vector machine (SVM) with a one-against-all approach.
• Validation using 10-fold cross-validation and leave-one-subject-out cross-validation on the sleep-EDF database.

Main Results:

• SC-En&SVM demonstrated high stability and effective performance.
• Average accuracies for 2-6 sleep states ranged from 97.02% ± 0.58 to 83.94% ± 1.61 (10-fold CV).
• Independent validation yielded accuracies of 94.15% to 75.98%, comparable or superior to state-of-the-art methods, with substantial agreement (kappa coefficients 0.67-0.81).

Conclusions:

• SC-En&SVM is a novel, simple, and efficient sleep stage scoring method.
• The method achieves high classification performance without compromising simplicity.
• This approach offers a valuable tool for healthcare applications requiring accurate sleep analysis.