Effect of EEG Dual-Channel Acquisition and Gender Specification Subjects on the Classification of Sleep Stages Using Machine Learning Techniques

Abstract

Nowadays sleep staging is considered one of the major issues in all age groups. Proper sleep scoring of sleep stages can give clinical information for selecting a suitable diagnosing of sleep disorder. Since the manual visual scoring of sleep stage classification is highly time-consuming and it also depends on expert experiences. To overcome this problem automatic sleep stage classification obtained to diagnosis sleep disorder. The main objective of this study is to automatic sleep stage classification on dual channels of EEG signals in gender-specific subjects. This study round up a wide range of research findings concerning sleep stage classification. Here we have combined the NREM and REM stage into one stage, called the sleep stage. Basically here we have proposed two-state classification and practically implemented on public available sleep dataset. In this study, we have considered two channels of EEG signals with different sex subjects. Feature selection and classifiers are assessed the accuracy level of which channel recordings and which classification algorithm is more suitable to discriminating the different sleep stages in subject-gender specific accurately. According to our achieved results from both the gender-specific subjects from dual channels of EEG signal, subject male category with F3-A2 channel and Ensemble classifier are chosen as the best channel and classifier. The mentioned channel and classifier have reached 90.7% accuracy, in the same manner for the female subject category with the C3-A2 channel and LSVM classifier are chosen as the best channel and classifier and have reached 91.5% accuracy levels in discriminating the different stages of sleep. This study shows that gender-specific and channel-specific recordings can be classifying sleep stages with a level accuracy that makes it more suitable for scientific and clinical sleep disorder assessment.