Sleep Stages Classification via EEG Signals Using Quadratic Support Vector Machine (SVM) Algorithm

Sleep stages classification plays a crucial role in diagnosing and understanding sleep disorders. With the increasing prevalence of sleep disorders, there is a growing need for automated methods to assist in sleep stage classification. In this study, we present a novel approach for sleep stages classification using EEG and machine learning models. Our method involves the extraction of EEG features from a dataset comprising recordings from six healthy patients. The feature extraction process includes Discrete Wavelet Transform (DWT) for sub-band extraction and the calculation of power spectral density (PSD), spectral features, Hjorth parameters, Hurst exponent, wavelet entropy, and autoregressive modeling. These features capture different aspects of the EEG signals, such as power distribution, waveform characteristics, complexity, and self-similarity. We train a Quadratic Support Vector Machine (SVM) classifier using the extracted features to classify sleep stages accurately. The Quadratic SVM algorithm allows for the optimization of a quadratic decision boundary, enabling the capture of complex relationships and nonlinear patterns in the data. The trained SVM-based model achieves a high testing accuracy of 97% in sleep stage classification.

I. Introduction

Sleep is a vital biological process that plays a critical role in maintaining overall health and well-being. However, sleep disorders, such as insomnia and sleep apnea, are prevalent and can have significant negative impacts on quality of life and overall health. The accurate classification of sleep stages is essential for diagnosing and understanding sleep disorders, as different stages have distinct patterns of brain activity and physiological responses.