DeepSleepNet: A Model for Automatic Sleep Stage Scoring Based on Raw Single-Channel EEG

Abstract

This paper proposes a deep learning model, named DeepSleepNet, for automatic sleep stage scoring based on raw single-channel EEG. Most of the existing methods rely on hand-engineered features, which require prior knowledge of sleep analysis. Only a few of them encode the temporal information, such as transition rules, which is important for identifying the next sleep stages, into the extracted features. In the proposed model, we utilize convolutional neural networks to extract time-invariant features, and bidirectional-long short-term memory to learn transition rules among sleep stages automatically from EEG epochs. We implement a two-step training algorithm to train our model efficiently. We evaluated our model using different single-channel EEGs (F4-EOG (left), Fpz-Cz, and Pz-Oz) from two public sleep data sets, that have different properties (e.g., sampling rate) and scoring standards (AASM and R&K). The results showed that our model achieved similar overall accuracy and macro F1-score (MASS: 86.2%-81.7, Sleep-EDF: 82.0%-76.9) compared with the state-of-the-art methods (MASS: 85.9%-80.5, Sleep-EDF: 78.9%-73.7) on both data sets. This demonstrated that, without changing the model architecture and the training algorithm, our model could automatically learn features for sleep stage scoring from different raw single-channel EEGs from different data sets without utilizing any hand-engineered features.

V‐Awake: A Visual Analytics Approach for Correcting Sleep Predictions from Deep Learning Models

“…To introduce our approach, we use a DL model [SDWG17] that scores sleep data. A graphical, high level description is shown in Fig.…”

“…In this model, convolutional layers act as feature extractors directly from the raw input signal, while LSTM layers learn transition rules between sleep stages. The model achieves an accuracy of 82.0% [SDWG17].…”

“…Deep learning model for sleep stage scoring [SDWG17]. It comprises two convolutional branches with different kernel sizes, a shortcut connection and two bidirectional LSTM layers.…”

“…This manual approach is time consuming and labor intensive [BKS∗17], making it hard to apply at a large scale. To overcome this limitation, a lot of research has been performed to automate sleep scoring tasks, for example, by using DL models [SDWG17, BKS∗17, LKL12, ZWBC16, TMGZ16]. The automation of the scoring process has obvious benefits regarding time and effort.…”

“…We conducted our research with a sleep scoring model trained on raw, single EEG channel data [SDWG17]. We demonstrate the usability of our approach in a concrete, real‐life use case together with two somnologists.…”