A comparative EEG analysis of classifying short sleep phases and waking states using support vector machine and random forest

Abstract Multiple studies suggest that the various stages of sleep affect the effectiveness of taking a nap. For this reason, the purpose of this study is to develop a model that may be used to classify the first and second stages of short sleep or the awake state. We employ sleep recordings obtained from the open-access dataset. To enhance the quality of recorded EEG signals, we implement a Notch Filter to reduce power line noise and a 0.5–70 Hz bandpass (Butterworth) filter to isolate the pertinent EEG signals. Two classifiers, Support Vector Machine (SVM) and Random Forest (RF), are used to assess and compare the performance of classification. In addition, the mRmR (minimal Redundancy Maximum Relevance) feature selection approach is employed to improve the model efficiency. The outcomes of our study reveal that both classifiers for each subject have an accuracy rate approaching 80%, differentiating between wakefulness and phases 1 and 2 of short sleep. This study emphasizes the efficacy of these strategies in offering essential instruments for comprehending and enhancing nap efficiency.