Expert Knowledge Inspired Contrastive Learning for Sleep Staging

Environmental requirement:

wget==3.2
pyedflib==0.1.22
torch==1.6.0+cu101
scikit-learn==0.24.1
scipy==1.5.4
numpy==1.19.5

Run the code:

Download the third-party public data set sleep-EDF:

python download_sleepedf.py

Prepare the single-channel EEG data:

python prepare_sleepedf.py

Extract prior features for self-supervised pretraining:

python prepare_feature.py

Run SleepECL:

python main.py --window_size 9 --sub_window_size 3 --cnn_level dcc prior --sa_level dcc prior --data_dir ./data/sleepedf/sleep-cassette/eeg_fpz_cz --feature_path ./data/sleepedf/sleep-cassette/feature/eeg_fpz_cz_powerband.pkl