An intelligent system for diagnosing sleep stages using wavelet coefficients

Abstract

Human sleep is divided into two segments, Rapid Eye Movement (REM) sleep and Non-REM (NREM) sleep. NREM sleep is further divided into 4 stages. Sleep staging attempts to identify these stages based on the signals collected in PSG. Significant information can be derived from the EEG signals collected during PSG. Wavelet coefficients are extracted from EEG signals. In order to reduce the amount of data set, the statistical features are calculated from wavelet coefficients. For performing decision making, six ANFIS classifiers and SVM classifier are used to differentiate between REM and Non-REM sleep stages. That is to say, pattern varies under the different sleep stages. Therefore, healthy humans with a regular night's sleep will follow these sleep stages in a particular pattern.