Mobile QR Code

1. (Department of Computer Engineering, Kwangwoon University / Seoul, Korea {zhsjzhsj, jw03070, hyunsoo940324, cndtjq97}@gmail.com, parkcheolsoo@kw.ac.kr )

Sleep stage, Attention, Explainable artificial intelligence, Electroencephalogram, Signal decompose

## 1. Introduction

Recent studies have shown that sleep disorder patients continue to increase [1], and people with sleep disorders are more vulnerable to various motor diseases [1] or disorders such as Alzheimer's disease [3,4]. The most important part for the treatment of sleep disorders is estimation of the sleep stages. Based on the sleep stages, a doctor can determine the overall sleep quality and amount and then conduct a sleep diagnosis based on the analysis. However, the estimation of sleep stage can only proceed with the approach of manual sleep estimation by sleep specialists at a medical site [5], which results in high costs of economic resources and time. In addition, recent research reports that sleep experts’ sleep stage estimation results vary wildly from one to another [6] since the sleep stage is determined by their subjective judgment.

To solve these problems, various methods of estimating the sleep stages using neural networks have recently been proposed. Supratak et al. [7] proposed a sleep stage automation architecture using convolution network layers [8] and long short-term memory (LSTM) cells [8]. However, there is a black-box problem of not being able to explain the resulting process due to the use of a deep neural network [9]. Results reported for 21 limited subjects have led to insufficient generalization performance and validity of the model.

In 2019, Mousavi et al. [10] proposed a sleep stage estimation model using a convolutional neural network (CNN) and bidirectional recurrent neural network (Bi-RNN) [8] for seq2seq learning [11,16]. The results showed a slight improvement in accuracy compared to a previously proposed method, but the problem of using only a small number of subjects was not solved. Also, the black-box problem was further complicated because they used multiple Bi-RNNs.

The medical community is increasingly demanding explainable artificial intelligence (XAI) [12,13]. Thus, an explainable artificial neural network model should be proposed for medical decision support systems. However, no previously proposed neural networks [7,10] at an explainable level have been proposed for sleep automation models through artificial neural network algorithms.

In this study, we propose a deep neural network-based explainable model that estimates five sleep stages based on an EEG band component. The attention layer in the model updates its attention weights to learn which EEG band components are useful for an estimate of a specific sleep stage. Through these methods, the model learns from multiple -decomposed EEG band components, which sleep specialists suggest may help to estimate more accurate sleep stages [27,30]. In addition to these proposals, we demonstrate the generalization performance and increase the robustness of the model by using at least five times more subject data (80 subjects) than previous studies [7,10].

## 2. Methodology

### 2.1 Training Data

A previously proposed sleep stage estimation model has been taught primarily using EEG signals recorded in Fpz-Cz or Pz-Oz configurations [7,10]. However, the corresponding EEG channel is not easy to measure and is reported to be noisy [17]. In comparison, brain waves of F3-M2 channels measured by attaching electrodes directly to the epidermis have the advantage of relatively low noise and robustness to external signals. In this study, we used 5-level (Wake, REM, N1, N2, and N3) sleep stage labels that were estimated every 30 seconds. This was based on data from a single EEG channel for 80 participants’ data with the F3-M2 channel configuration among the EEG data measured in polysomnography (PSG) in the Wisconsin Sleep Cohort sleep dataset [18,19]. To eliminate unnecessary ambient noise in the F3-M2-channel EEG signal, a bandpass filter with a 500 orders in a frequency range between 0.5 Hz and 55 Hz was applied [20]. Additionally, notch filters were applied at 60 Hz to reduce the multiple powerline noise [21].

### 2.2 Signal Decomposition

In order to decompose an EEG signal, many decomposition methods have been proposed, such as fast Fourier transform (FFT) [22] and wavelet transform (WT) methods [23,24]. In this study, to obtain the weight for a particular EEG band component, a signal decomposition through an FIR band-pass filter was applied to a single-channel EEG to train a model based on conventional knowledge of sleep staging reported by sleep experts. This band-pass filter is a zero-phase band-pass filter with 500 orders. A different band-pass filter was designed for each EEG frequency component.

The beta frequency component was extracted from a single-channel EEG at 13-38 Hz, the alpha component was extracted at 8-12 Hz, the theta component was extracted at 4-7 Hz, and the delta component was extracted at 1-3 Hz [14]. We used these extracted brainwave components to estimate sleep stages, which have been used by sleep experts as well [27-30]. Model parameters were then updated through these decomposed signal data.

### 2.2 Model Specifications

The attention layer [16] is located before the overall input of the model architecture, as shown in Fig. 1. How much each signal component is learned is reflected in the final label, which ranges from -1.0 to +1.0. The model can analyze the attention weight for each post-learning estimation stage by the attention layer structure. After learning the input data, analysis of the channel weights of the relevant attention layer allows us to analyze which EEG frequency component the model relies on more for estimating specific sleep stages.

To predict the sleep stage through the EEG data, the model performs representation learning and temporal learning. In the representation learning parts, 30 seconds of decomposed EEG signals were given as input to the 1D convolutional neural network (CNN). Small neural networks for each EEG frequency component were designed and constructed by a 1D CNN [8] while pooling layers eight times. Based on this structure, features for each EEG component can be learned appropriately for the input data for each CNN structure.

To learn about the time series elements of EEG data, temporal learning is proposed through Bi-GRU. Bi-GRU enables the learning of an epoch time unit with every 10 time steps for single-channel EEG information for 30 seconds of a sleep EEG signal. A model with this structure can learn sequential information in the time domain and extract features from the input EEG frequency components. The model hyper parameter is optimized based on previous sleep stage estimation algorithms [7,10], and parameters are updated by an adaptive moment estimation (ADAM) optimizer [8].

## 3. Results

To evaluate the proposed model, performance measurement of the model was performed using the F1-score [26], the most used indicator in sleep stage analysis [7,10]. The results of 5-fold cross-validation [25] of the sleep stage estimation model for 80 subjects’ single-channel EEG sleep data with the proposed method are shown in Table 2. Cross-validation of the model showed that the average F1-score was 70.61% (${\pm}$26.34).

To analyze the major EEG components from the EEG decomposition using the model in estimating sleep stages, attention weights were used to calculate which EEG frequency components have more weights to track their update rate for each sleep stage. The attention-weight analysis demonstrates the channel-specific attention weights used to estimate each sleep stage among the five sleep-stage estimation labels, as shown in Fig. 2. The results in Fig. 2 confirm the previous findings that different valid EEG signal components are used to estimate the sleep stages [27-30]. In light sleep, such as N1 and N2, and REM, high-frequency components have more attention rates, as shown in Fig. 2. This shows that the model estimates the sleep stage mainly based on the high-frequency components, whereas deep sleep (N3) shows more attention weight in the low-frequency component, showing that deep sleep is estimated mainly based on the low-frequency component.

## 4. Discussion

Analysis of the attention weights of the proposed model confirms that the model learns through different channel weights corresponding to different EEG components. This aligns with previous knowledge for sleep stage estimation that different brainwave components are used for estimating different sleep stages [28-30]. The proposed model has more clearly described which component of the EEG is more useful to detect specific sleep stages based on the selective focus learned from different frequency components of the input data to estimate a particular sleep stage. In addition, the results for the attention weights of each EEG component show similar results to those of useful EEG components reported in previous papers analyzing sleep stage through EEG channels. This shows the model might learn how to evaluate the sleep stage around the low-frequency component for deep sleep and the high-frequency component for light sleep, similar to what sleep specialists do for their estimation of the sleep stages.

Furthermore, considering the characteristics of sleep-stage estimation models, which show worse performance as the number of subjects increases, the model still shows similar F1-scores, even when subject numbers increase by more than 5 times compared to previously reported sleep-stage estimation algorithms. This implies that attention mechanisms in sleep stage analysis could be effective in estimating the sleep stages and designing an explainable model.

## 5. Conclusion

In this paper, we proposed an explainable artificial intelligence model centered on brainwave components through attention algorithms. Based on that model, we were able to show which brainwave components were useful in predicting specific sleep stages, and the accuracy was also similar to that of previously proposed sleep stage estimation models. However, the problem of chronic sleep stage estimation bias caused by an imbalance of the sleep stage can be still observed. This is particularly shown in the N1 sleep stage.

To solve this problem, we propose increasing the learning volume through the synthetic minority oversampling technique (SMOTE) [31] and data augmentation [32] to improve the performance of estimating the N1 sleep stage. In addition, there are still some black-box problems because the level of the explainable model cannot show the learning progress by time frame. To solve this problem, we propose a study of a deeper-level explainable model through time-unit attention [33].

### ACKNOWLEDGMENTS

This research was supported by the MIST(Ministry of Science and ICT), under the National Program for Excellence in SW (2017-0-00096), supervised by the IITP(Institute for Information & communications Technology Promotion )

### REFERENCES

1
Chokroverty, Sudhansu. , 1999, Overview of sleep & sleep disorders., Indian J Med Res, Vol. 131, No. 2, pp. 126-140
2
Factor S. A., McAlarney T., Sanchez‐Ramos J. R., Weiner W. J., 1990, Sleep disorders and sleep effect in Parkinson’s disease, Mov. Disord., Vol. 5, No. 4, pp. 280-285
3
Ju Y.-E. S., Lucey B. P., Holtzman D. M., 2013, Sleep and alzheimer disease pathology-a bidirectional relationship, Nature Reviews Neurology, Vol. 10, No. 2, pp. 115-119
4
Leng M., Yin H., Zhang P., Jia Y., Hu M., Li G., Wang C., Chen L., 2020, Sleep quality and health-related quality of life in older people with subjective cognitive decline, mild cognitive impairment, alzheimer disease, Journal of Nervous & Mental Disease, Vol. 208, No. 11, pp. 913-913
5
Mouthon A.-L., Huber R., 2015, Methods in pediatric sleep research and sleep medicine, Neuropediatrics, Vol. 46, No. 03, pp. 159-170
6
Whitney C. W., Gottlieb D. J., Redline S., Norman R. G., Dodge R. R., Shahar E., Surovec S., Nieto F. J., 1998, Reliability of scoring respiratory disturbance indices and sleep staging, Sleep, Vol. 21, No. 7, pp. 749-757
7
Supratak A., Dong H., Wu C., Guo Y., 2017, Deepsleepnet: A model for automatic sleep stage scoring based on raw single-channel eeg, IEEE Transactions on Neural Systems and Rehabilitation Engineering, Vol. 25, No. 11, pp. 1998-2008
8
Schmidhuber J., 2015, Deep learning in neural networks: An overview, Neural Networks, Vol. 61, pp. 85-117
9
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
10
Mousavi S., Afghah F., Acharya U. R., 2019, Sleepeegnet: Automated sleep stage scoring with sequence to sequence deep learning approach, PLOS ONE, Vol. 14
11
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
12
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
13
Pocevičiūtė M., Eilertsen G., Lundström C., 2020, Survey of Xai in digital pathology, Artificial Intelligence and Machine Learning for Digital Pathology, pp. 56-88
14
Daud S. S., Sudirman R., 2015, Butterworth bandpass and Stationary wavelet transform Filter comparison For Electroencephalography Signal, 2015 6th International Conference on Intelligent Systems, Modelling and Simulation
15
Hassan A. R., Subasi A., 2017, A decision support system for automated identification of sleep stages from SINGLE-CHANNEL EEG signals, Knowledge-Based Systems, Vol. 128, pp. 115-124
16
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
17
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
18
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
19
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
20
Buhrmester V., Münch D., Arens M., 2019, Analysis of Explainers of Black Box Deep Neural Networks for Computer Vision: A Survey, no. Ml, pp. 1-22
21
Tibdewal M. N., Mahadevappa M., Ray A. K., Malokar M., Dey H. R., 2016, Power line and ocular artifact denoising from EEG using notch filter and wavelet transform, Proc. 10th INDIACom; 2016 3rd Int. Conf. Comput. Sustain. Glob. Dev. INDIACom 2016, pp. 1654-1659
22
Schirm L., 1979, Fast Fourier Transforms., New Electron, Vol. 12, No. 2
23
Arneodo A., Grasseau G., Holschneider M., 1988, Wavelet transform of multifractals, Physical Review Letters, Vol. 61, No. 20, pp. 2281-2284
24
Kevric J., Subasi A., 2017, Comparison of signal decomposition methods in classification of EEG signals for motor-imagery BCI system, Biomed. Signal Process. Control, Vol. 31, pp. 398-406
25
Hastie T., Friedman J., Tisbshirani R., 2017, The elements of statistical learning: Data mining, inference, prediction., New York: Springer
26
Sokolova M., Japkowicz N., Szpakowicz S., 2006, Beyond accuracy, F-score and ROC: A family of discriminant measures for performance evaluation, AAAI Work. - Tech. Rep., Vol. ws-06-06, pp. 24-29
27
Sokolova M., Japkowicz N., Szpakowicz S., 2006, Beyond accuracy, F-score and ROC: A family of discriminant measures for performance evaluation, AAAI Work. - Tech. Rep., Vol. ws-06-06, pp. 24-29
28
Sokolova M., Japkowicz N., Szpakowicz S., 2006, Beyond accuracy, F-score and ROC: A family of discriminant measures for performance evaluation, AAAI Work. - Tech. Rep., Vol. ws-06-06, pp. 24-29
29
Abeln V., Kleinert J., Strüder H. K., Schneider S., 2014, Brainwave entrainment for better sleep and post-sleep state of young elite soccer players - A pilot study, Eur. J. Sport Sci., Vol. 14, No. 5, pp. 393-402
30
Abeln V., Kleinert J., Strüder H. K., Schneider S., 2014, Brainwave entrainment for better sleep and post-sleep state of young elite soccer players - A pilot study, Eur. J. Sport Sci., Vol. 14, No. 5, pp. 393-402
31
Kovács B., Tinya F., Németh C., Ódor P., 2020, Unfolding the effects of different forestry treatments on microclimate in oak forests: results of a 4-yr experiment, Ecol. Appl., Vol. 30, No. 2, pp. 321-357
32
Wang F., Zhong S. H., Peng J., Jiang J., Liu Y., 2018, Data augmentation for eeg-based emotion recognition with deep convolutional neural networks, Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), Vol. 10705 LNCS, pp. 82-93

## Author

##### Suwhan Baek

Suwhan Baek received his B.S. degree in Computer Engineering from Kwangwoon University in Seoul, South Korea. His research interests include overall Medical AI and Auto ML(ENAS). He is also attracted by reinforcement learning, generative models.

##### Jaewoo Baek

Jaewoo Baek is in the MSc Program at the Bio Computing & Machine Learning Laboratory (BCML) in the Department of Computer Engineering at Kwangwoon University, Seoul, Republic of Korea. His research interests include biological signal processing, machine learning, deep learning, and reinforcement learning

##### Hyunsoo Yu

Hyunsoo Yu received his BS degree in Robotics Engineering from Kwangwoon University in Seoul, South Korea. His reasearch interests include in experimental setting, signal processing, machine learning and artificial Intelligence.

##### Choongseop Lee

Choongseop Lee received his BS degree in Computer Software and Engineering from Kwangwoon University in Seoul, South Korea. His reasearch interests include machine learning and computational neuroscience.

##### Cheolsoo Park

Cheolsoo Park is an associate professor in the Computer Engineering Department at Kwangwoon University, Seoul, South Korea. He received a B.Eng. in Electrical Engineering from Sogang University, Seoul, and an MSc from the Biomedical Engineering Department at Seoul National University, South Korea. In 2012, he received his PhD in Adaptive Nonlinear Signal Processing from Imperial College London, London, U.K., and worked as a postdoctoral researcher in the Bioengineering Department at the University of California, San Diego, U.S.A. His research interests are mainly in the areas of machine learning and adaptive and statistical signal processing, with applications in brain computer interfaces, computational neuroscience, and wearable technology.