TinySleepNet

An efficient deep learning model for sleep stage scoring based on raw, single-channel EEG. This work is an extension of our previous work, DeepSleepNet [paper][github].

Code for the model in the paper TinySleepNet: An Efficient Deep Learning Model for Sleep Stage Scoring based on Raw Single-Channel EEG by Akara Supratak and Yike Guo from The Faculty of ICT, Mahidol University and Imperial College London respectively.

This work has been accepted for publication in 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC).

This work has been used as a baseline for quantifying the transferability in comparison with U-Time model. The results have been published in the paper Quantifying the impact of data characteristics on the transferability of sleep stage scoring models by Akara Supratak and Peter Haddawy from The Faculty of ICT, Mahidol University [paper][github].

Model Architecture

Note: Fs is the sampling rate of the input EEG signals

Performance Comparison

Note: ACC = accuracy, MF1 = Macro F1-Score

Environment

• CUDA 10.0
• cuDNN 7
• Tensorflow 1.13.1

Create a virtual environment with conda

conda create -n tinysleepnet python=3.6
conda activate tinysleepnet
pip install -r requirements.txt

How to run

1. python download_sleepedf.py
2. python prepare_sleepedf.py
3. python trainer.py --db sleepedf --gpu 0 --from_fold 0 --to_fold 19
4. python predict.py --config_file config/sleepedf.py --model_dir out_sleepedf/train --output_dir out_sleepedf/predict --log_file out_sleepedf/predict.log --use-best

Citation

If you find this useful, please cite our work as follows:

@INPROCEEDINGS{Supratak2020,
    title = {TinySleepNet: An Efficient Deep Learning Model for Sleep Stage Scoring based on Raw Single-Channel EEG},
    author = {Supratak, Akara and Guo, Yike},
    booktitle = {2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC),
    year = {2020},
    volume = {}, 
    number = {}, 
    pages = {641-644}, 
    doi = {10.1109/EMBC44109.2020.9176741}, 
    ISSN = {}, 
}

Licence

• For academic and non-commercial use only
• Apache License 2.0