A CNN-LSTM Model for Sleep Stage Scoring Using EEG Signals

This paper proposes a deep-learning model for automated sleep stage scoring with single-channel EEG recordings that combines CNN and LSTM. Several of the methods in use today rely on qualities that have been hand-engineered. However, only a few techniques retrieve the temporal data necessary for determining the phases of sleep. The extraction of time-invariant features will be possible using convolutional neural networks, and LSTM will be able to learn the transition rules. The model was trained and tested using the Sleep-EDF-v1 dataset, which is publicly available. We used single-channel Fpz-Cz EEGs and scored them in accordance with the AASM standard. 83.38% accuracy was achieved overall.