In recent years, deep-learning models gained attention for electroencephalography (EEG) classification tasks due to their excellent performance and ability to extract complex features from raw data. In particular, convolutional neural networks (CNN) showed adequate results in brain-computer interfaces (BCI) based on different control signals, including event-related potentials (ERP). In this study, we propose a novel CNN, called EEG-Inception, that improves the accuracy and calibration time of assistive ERP-based BCIs. To the best of our knowledge, EEG-Inception is the first model to integrate Inception modules for ERP detection, which combined efficiently with other structures in a light architecture, improved the performance of our approach. The model was validated in a population of 73 subjects, of which 31 present motor disabilities. Results show that EEG-Inception outperforms 5 previous approaches, yielding significant improvements for command decoding accuracy up to 16.0%, 10.7%, 7.2%, 5.7% and 5.1% in comparison to rLDA, xDAWN + Riemannian geometry, CNN-BLSTM, DeepConvNet and EEGNet, respectively. Moreover, EEG-Inception requires very few calibration trials to achieve state-of-the-art performances taking advantage of a novel training strategy that combines cross-subject transfer learning and fine-tuning to increase the feasibility of this approach for practical use in assistive applications.

Rationale: The vast majority of children around the world undergoing adenotonsillectomy for obstructive sleep apnea–hypopnea syndrome (OSA) are not objectively diagnosed by nocturnal polysomnography because of access availability and cost issues. Automated analysis of nocturnal oximetry (nSpO₂), which is readily and globally available, could potentially provide a reliable and convenient diagnostic approach for pediatric OSA. Methods: Deidentified nSpO₂ recordings from a total of 4,191 children originating from 13 pediatric sleep laboratories around the world were prospectively evaluated after developing and validating an automated neural network algorithm using an initial set of single-channel nSpO₂ recordings from 589 patients referred for suspected OSA. Measurements and Main Results: The automatically estimated apnea–hypopnea index (AHI) showed high agreement with AHI from conventional polysomnography (intraclass correlation coefficient, 0.785) when tested in 3,602 additional subjects. Further assessment on the widely used AHI cutoff points of 1, 5, and 10 events/h revealed an incremental diagnostic ability (75.2, 81.7, and 90.2% accuracy; 0.788, 0.854, and 0.913 area under the receiver operating characteristic curve, respectively). Conclusions: Neural network–based automated analyses of nSpO₂ recordings provide accurate identification of OSA severity among habitually snoring children with a high pretest probability of OSA. Thus, nocturnal oximetry may enable a simple and effective diagnostic alternative to nocturnal polysomnography, leading to more timely interventions and potentially improved outcomes.

This study aims at assessing the usefulness of deep learning to enhance the diagnostic ability of oximetry in the context of automated detection of pediatric obstructive sleep apnea (OSA). A total of 3196 blood oxygen saturation (SpO ₂ ) signals from children were used for this purpose. A convolutional neural network (CNN) architecture was trained using 20-min SpO ₂ segments from the training set (859 subjects) to estimate the number of apneic events. CNN hyperparameters were tuned using Bayesian optimization in the validation set (1402 subjects). This model was applied to three test sets composed of 312, 392, and 231 subjects from three independent databases, in which the apnea-hypopnea index (AHI) estimated for each subject (AHI _CNN ) was obtained by aggregating the output of the CNN for each 20-min SpO ₂ segment. AHI _CNN outperformed the 3% oxygen desaturation index (ODI3), a clinical approach, as well as the AHI estimated by a conventional feature-engineering approach based on multi-layer perceptron (AHI _MLP ). Specifically, AHI _CNN reached higher four-class Cohen's kappa in the three test databases than ODI3 (0.515 vs 0.417, 0.422 vs 0.372, and 0.423 vs 0.369) and AHI _MLP (0.515 vs 0.377, 0.422 vs 0.381, and 0.423 vs 0.306). In addition, our proposal outperformed state-of-the-art studies, particularly for the AHI severity cutoffs of 5 e/h and 10 e/h. This suggests that the information automatically learned from the SpO ₂ signal by deep-learning techniques helps to enhance the diagnostic ability of oximetry in the context of pediatric OSA.

Automatic deep-learning models used for sleep scoring in children with obstructive sleep apnea (OSA) are perceived as black boxes, limiting their implementation in clinical settings. Accordingly, we aimed to develop an accurate and interpretable deep-learning model for sleep staging in children using single-channel electroencephalogram (EEG) recordings. We used EEG signals from the Childhood Adenotonsillectomy Trial (CHAT) dataset (n = 1637) and a clinical sleep database (n = 980). Three distinct deep-learning architectures were explored to automatically classify sleep stages from a single-channel EEG data. Gradient-weighted Class Activation Mapping (Grad-CAM), an explainable artificial intelligence (XAI) algorithm, was then applied to provide an interpretation of the singular EEG patterns contributing to each predicted sleep stage. Among the tested architectures, a standard convolutional neural network (CNN) demonstrated the highest performance for automated sleep stage detection in the CHAT test set (accuracy = 86.9% and five-class kappa = 0.827). Furthermore, the CNN-based estimation of total sleep time exhibited strong agreement in the clinical dataset (intra-class correlation coefficient = 0.772). Our XAI approach using Grad-CAM effectively highlighted the EEG features associated with each sleep stage, emphasizing their influence on the CNN's decision-making process in both datasets. Grad-CAM heatmaps also allowed to identify and analyze epochs within a recording with a highly likelihood to be misclassified, revealing mixed features from different sleep stages within these epochs. Finally, Grad-CAM heatmaps unveiled novel features contributing to sleep scoring using a single EEG channel. Consequently, integrating an explainable CNN-based deep-learning model in the clinical environment could enable automatic sleep staging in pediatric sleep apnea tests.

Patients with obstructive sleep apnea (OSA) exhibit heterogeneous heart rate variability (HRV) during wakefulness and sleep. We investigated the influence of OSA severity on HRV parameters during wakefulness in a large international clinical sample. 1247 subjects (426 without OSA and 821 patients with OSA) were enrolled from the Sleep Apnea Global Interdisciplinary Consortium. HRV parameters were calculated during a 5-minute wakefulness period with spontaneous breathing prior to the sleep study, using time-domain, frequency-domain and nonlinear methods. Differences in HRV were evaluated among groups using analysis of covariance, controlling for relevant covariates. Patients with OSA showed significantly lower time-domain variations and less complexity of heartbeats compared to individuals without OSA. Those with severe OSA had remarkably reduced HRV compared to all other groups. Compared to non-OSA patients, those with severe OSA had lower HRV based on SDNN (adjusted mean: 37.4 vs. 46.2 ms; p < 0.0001), RMSSD (21.5 vs. 27.9 ms; p < 0.0001), ShanEn (1.83 vs. 2.01; p < 0.0001), and Forbword (36.7 vs. 33.0; p = 0.0001). While no differences were found in frequency-domain measures overall, among obese patients there was a shift to sympathetic dominance in severe OSA, with a higher LF/HF ratio compared to obese non-OSA patients (4.2 vs. 2.7; p = 0.009). Time-domain and nonlinear HRV measures during wakefulness are associated with OSA severity, with severe patients having remarkably reduced and less complex HRV. Frequency-domain measures show a shift to sympathetic dominance only in obese OSA patients. Thus, HRV during wakefulness could provide additional information about cardiovascular physiology in OSA patients. A Prospective Observational Cohort to Study the Genetics of Obstructive Sleep Apnea and Associated Co-Morbidities (German Clinical Trials Register - DKRS, DRKS00003966) https://www.drks.de/drks_web/navigate.do?navigationId=trial.HTML&TRIAL_ID=DRKS00003966