Deep Learning-Based Sleep Stage Classification Using EEG

Classification of sleep stages is an essential area of research that helps develop treatments for people with sleep disorders. According to common sleep stage criteria, sleep is divided into six different stages: Wakeful sleep (W), REM (rapid eye movement) sleep, and non-REM sleep (S1-S4). Sleep processing can be performed by analyzing electroencephalogram (EEG) signals in a 30-second cycle (epoch). These stages are chosen and established on an analysis of brain workouts during sleep. This reveals a clear pattern that characterizes each stage. Sleep deprivation can cause various illnesses, including obesity, heart disease, diabetes, and reduced life expectancy [2]. Sleep professionals usually classify sleep stages into polysomnography (PSG) signals. Polysomnography consists of an electroencephalogram (EEG), electro-oculogram (EOG), electromyogram (EMG), and electrocardiogram (ECG) [2]. In addition, one category of such classifiers, Deep Learning (DL) based EEG signal classification, is used to classify sleep stages. The treatise includes an analysis of the performance of the considered methods of sleep grading. In addition, the strengths and weaknesses of classical and deep learning-based sleep staging methods will be explored. In addition, we compared standard classification with the data fusion methods with their accuracy.