TWNSleepNet: An Automatic Sleep Staging Model Based on Causal Expansion Convolutional and Residual Networks

The Electroencephalogram (EEG) is a widely used fundamental signal for monitoring brain activity and diagnosing sleep disorders. The manual staging of artificial sleep is a time-consuming and laborious task for sleep experts. To address this issue, this paper proposes an automatic sleep staging method based on single-channel EEG signals, named TWNSleepNet. TWNSleepNet comprises causal dilation convolution and residual networks and is trained on the Sleep EDF dataset. Experimental results demonstrate that our proposed model achieves outstanding performance in sleep staging tasks.