Automatic Sleep Stage Classification Using EEG and EMG Signal

Sleep is a primary constituent of human life. It is important to maintain good sleep efficiency because some problems occur when sleep efficiency is low. Sleep efficiency is calculated by the ratio of sleep stages. Sleep stages can be classified using Polysomnogram (PSG), which includes information of EEG, EMG and EOG. There have been many studies to classify sleep stages automatically using EEG signal. They, however, have difficulty in classifying several sleep stages because of the resemblance of EEG signals, especially, REM and Non-REM1 (N1) stage. We propose to use EMG signal in addition to EEG signal to improve the accuracy of sleep stage classification. EMG signal is useful for classifying REM stage and Non-REM stages. We propose a machine learning model of Support Vector Machine (SVM) using EEG and EMG signal. The proposed model shows higher classification rate for REM and N1 stage than EEG only model.