Automatic Sleep Stage Classification of Children with Sleep-Disordered Breathing Using the Modularized Network

Abstract

To develop an automatic sleep stage analysis model for children and evaluate the effect of the model on the diagnosis of sleep-disordered breathing (SDB). Patients and Methods: Three hundred and forty-four SDB patients aged between 2 to 18 years who completed polysomnography (PSG) to assess the severity of the disease were enrolled in this study. We developed deep neural networks to stage sleep from electroencephalography (EEG), electrooculography (EOG) and electromyogram (EMG). The model performance was estimated by accuracy, precision, recall, F1-score, and Cohen's Kappa coefficient (ĸ). And we compared the difference in calculation of sleep parameters among the technicians, the model ensemble, and the single-channel EEG model. Results:The numbers of raw data divided into training, validation, and testing were 240, 36, and 68, respectively. The best performance appeared in the model ensemble of which the accuracy was 83.36% (ĸ=0.7817) in 5-stages, and the accuracy was 96.76% (ĸ=0.8236) in 2-stages. The single-channel EEG model showed the classification satisfyingly as well. There was no significant difference in TST, SE, SOL, time in W, time in N1+N2, time in N3, and OAHI between technician and the model (P>0.05). On the datasets from sleep-EDF-13 and sleep-EDF-18, the average classification accuracies achieved were 92.76% and 91.94% in 5-stages by using the proposed method, respectively. Conclusion: This research established the model for pediatric automatic sleep stage classification with satisfying reliability and generalizability. In addition, it could be applied for calculating quantitative sleep parameters and evaluating the severity of SDB.

Deep learning-based algorithm accurately classifies sleep stages in preadolescent children with sleep-disordered breathing symptoms and age-matched controls

“…In comparison, our study achieved a similar performance with a larger pediatric cohort using only a 3-channel input. In parallel with the development of our work, other studies have also focused on sleep staging including pediatric patients and have demonstrated similar performance metrics (40,42,44). Notably, Wang et al (42) achieved high classification performance (with slightly lower kappa values compared to the present study when using a similar 3-channel input) with a modularized network utilizing a clinical pediatric dataset of 344 SDB patients with age 2-18 years (Table 5).…”

“…Numerous published studies have also attempted to fully automate the sleep staging process . Whilst historically, these have used feature engineering approaches or handcrafted rules (29-32), most recent studies utilize deep learning-based algorithms (22)(23)(24)(25)(26)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46). Although modern deep learning-based approaches generally perform well (kappa agreement typically ranging between 0.67 and 0.87) (48)(49)(50)(51), the majority have focused on adult populations (22, 23, 30-39, 41, 43, 45).…”

“…A smaller number of recent studies have focused on automatic sleep staging in children (24-29, 40, 42, 44, 47). Whilst some of these focus on two-or three-stage sleep classification (24,(26)(27)(28) [predominantly those considered infants (26)(27)(28)] or using non-EEGbased approaches intended for limited channel screening (29, 47), studies published in parallel with the development of this work using electrophysiological channels have demonstrated high sleep classification performance (40,42,44). However, there are some important limitations.…”

Influence of Channel Selection and Subject’s Age on the Performance of the Single Channel EEG-Based Automatic Sleep Staging Algorithms

“…In [ 27 ] a CNN-based sleep staging model has been applied to private data collected from 344 children aged 2–18 years old. Although the data they used has a wide age range, the model was trained and tested using data from all age ranges, and the reported performance was averaged over all participants due to the lack of sufficient participants at each age group to train age-specific models.…”

“…Very limited methods have been applied to children’s EEG signals. In [ 27 ], a deep neural network model was proposed for classifying children’s sleep stages. The proposed model utilized a modularized architecture that enables the neural network to have many layers without being constrained by the increasing number of hyperparameters.…”