Sleep stage classification is one of the critical methodologies for the diagnosis of sleep-related diseases and complications. The conventional method of categorization is quite clumsy and timeconsuming. This project aims to devise a deep learning and machine learning model for automatic classification of sleep stage, hence, removing the barrier of conventional method and expert ubiquity. In this work, we have considered a database that carries 197-night sleep polysomnographic data. Moreover, we aimed to classify this data into stages W, N1, N2, N3 and N4 as mentioned in the AASM standard. In addition to that, we have selected the EEG FpzCz channel because of its better quality and used an epoch time of 30 seconds for signal processing. We have used four machine learning and deep learning methods, namely CNN-CNN, CNN-LSTM, Random Forest, and XGBoosting, with 82%, 87%, 51%, and 59%, respectively. This report has depicted a roadmap of the EEG-based sleep stage scoring method by implementing the state of art methods. In conclusion, using better signal processing techniques will increase the overall performance and accuracy of the model.
The Dataset is available at https://drive.google.com/drive/folders/1Vkn-Yk2N8pF9rI0TFv7-pd9HvEXrtgz9