An Automated System for Sleep Staging using EEG Brain Signals Based on A Machine Learning Approach

Correctly classifying the sleep stages is essential for analyzing sleep quality and diagnosing sleep disorders. This research article aims to explore the classification performance of state-of-the-art machine learning (ML) models for classifying the sleep stages of different health-conditioned subjects and to identify whether the model performed well consistently for all kinds of sleep-disordered data. However, it has been noticed that the existing sleep staging methods have the following limitations:(1) Manual inspection of sleep behavior requires experts’ knowledge and which requires more workforce and a time-consuming process (2) Due to similar characteristics of the sleep stages features, it is essential to discriminate the changes aspects of the individual sleep stages with proper selection of parts (3) Acquisition of different medical-conditioned data has high demands on equipment. To address these challenges, we proposed an ML-based automated sleep staging system to analyze and classify sleep patterns. We use three different subgroups, namely SG-I, SG-II, and SG-III, to validate the proposed model’s effectiveness. First, our proposed model considered three different categories of sleep data with varying recordings of sessions and then extracted various features to understand the sleep behavior and applied feature selection algorithms to validate the feature concerning different subgroups’ sleep dates. Finally, apply the support vector machine (SVM) algorithm to classify them as Wake vs. Sleep on different sub-groups data ISRUC-Sleep dataset. The experimental results of the proposed method on the public repository ISRUC-Sleep show that it is improved. The results signify that the proposed method performs better than a state-of-the-art method.