An Intelligent Framework for Sleep Stage Scoring by 1D-Convolutional Neural Network with Bi-LSTM Layer Using EEG signal

To properly identify and treat sleep disorders, the stages of sleep must be successfully classified.Both manual and automatic sleep-scoring processes may be affected by such problems, but automatic techniques are more likely to be affected.Electroencephalogram (EEG) is a key factor in diagnosing sleep disorders instead of other biological signs.Historically, this time-consuming, arduous, and subject to subjective mistake task has been carried out manually by professionals through thorough observation.This process can be made significantly easier with automatic sleep "stage scoring" that performs scoring manually, making monitoring sleep disorders at work easier.A novel efficient "sleep stage Scoring" model is proposed using a deep learning approach to meet the requirement.The proposed model comprises four phases: Signal collection, decomposition, Dimensionality reduction and Scoring.Firstly, the raw EEG signals are collected via standard datasets.It is further used for signal decomposition, achieved by Empirical Mode Decomposition (EMD).Further, the dimensionality is reduced by Principal Component Analysis (PCA).Finally, this reduced decomposed signal is fed into the Scoring model as a "Serial Cascaded One-Dimensional Convolutional Neural Network with Bidirectional Long Short-Term Memory (SC1D-BiLSTM) layer".Using this proposed methodology, an average accuracy of 93.17% is achieved.Various performance metrics are utilized to assess the model, and the values are presented in the results section.The datasets used in this study are the Sleep-EDF Database Expanded v1.0.0 and St. Vincent's University Hospital / University College Dublin Sleep Apnea Database from Physionet, and the results are presented in the results section.