Purpose: To develop an automated framework for sleep stage scoring from PSG via a deep neural network. Methods: An automated deep neural network was proposed by using a multi-model integration strategy with multiple signal channels as input. All of the data were collected from one single medical center from July 2017 to April 2019. Model performance was evaluated by overall classification accuracy, precision, recall, weighted F1 score, and Cohen’s Kappa. Results: Two hundred ninety-four sleep studies were included in this study; 122 composed the training dataset, 20 composed the validation dataset, and 152 were used in the testing dataset. The network achieved human-level annotation performance with an average accuracy of 0.8181, weighted F1 score of 0.8150, and Cohen’s Kappa of 0.7276. Top-2 accuracy (the proportion of test samples for which the true label is among the two most probable labels given by the model) was significantly improved compared to the overall classification accuracy, with the average being 0.9602. The number of arousals affected the model’s performance. Conclusion: This research provides a robust and reliable model with the inter-rater agreement nearing that of human experts. Determining the most appropriate evaluation parameters for sleep staging is a direction for future research.
Purpose: This study aimed to propose a novel deep-learning method for automatic sleep apneic event detection and thus to estimate the apnea hypopnea index (AHI) and identify obstructive sleep apnea (OSA) in an event-by-event manner solely based on sleep sounds obtained by a noncontact audio recorder. Methods: We conducted a cross-sectional study of participants with habitual snoring or heavy breathing sounds during sleep to train and test a deep convolutional neural network named OSAnet for the detection of OSA based on sleep sounds. Polysomnography (PSG) was conducted, and sleep sounds were recorded simultaneously in a regular room without noise attenuation. The study was conducted in two phases. In phase one, eligible participants were enrolled and randomly allocated into training and validation groups for deep learning algorithm development. In phase two, eligible patients were enrolled in a test group for algorithm assessment. Sensitivity, specificity, accuracy, unweighted Cohen kappa coefficient (κ) and the area under the curve (AUC) were calculated using PSG as the reference standard. Results: A total of 135 participants were randomly divided into a training group (n, 116) and a validation group (n, 19). An independent test group of 59 participants was subsequently enrolled. Our algorithm achieved a precision of 0.81 and sensitivity of 0.78 in the test group for overall sleep event detection. The algorithm exhibited robust diagnostic performance to identify severe cases with a sensitivity of 95.6% and specificity of 91.6%. Conclusion: Our results showed that a deep learning algorithm based on sleep sounds recorded by a noncontact voice recorder served as a feasible tool for apneic event detection and OSA identification. This technique may hold promise for OSA assessment in the community in a relatively comfortable and low-cost manner. Further studies to develop a tool based on a home-based setting are warranted.
The deployment of learning-based models to detect malicious activities in network traffic flows is significantly challenged by concept drift. With evolving attack technology and dynamic attack behaviors, the underlying data distribution of recently arrived traffic flows deviates from historical empirical distributions over time. Existing approaches depend on a significant amount of labeled drifting samples to facilitate the deep model to handle concept drift, which faces labor-intensive manual labeling and the risk of label noise. In this paper, we propose ReCDA, a Concept Drift Adaptation method with Representation enhancement, which consists of a self-supervised representation enhancement stage and a weakly-supervised classifier tuning stage. Specifically, in the initial stage, ReCDA introduces drift-aware perturbation and representation alignment to facilitate the model in acquiring robust representations from drift-aware and drift-invariant perspectives. Moreover, in the subsequent stage, a meticulously crafted instructive sampling strategy and a robust representation constraint encourage the model to learn discriminative knowledge about benign and malicious activities during fine-tuning, thereby enhancing performance further. We conduct comprehensive evaluations on several benchmark datasets under varying degrees of concept drift. The experiment results demonstrate the superior adaptability and robustness of the proposed method.
To protect the information of video stream, many selective video encryption schemes have been proposed based on the H.265/HEVC video. However, most of the existing algorithms are not robust, thus failing to decrypt under packet loss. To further improve the robustness capability of video protection, a robust selective encryption scheme is proposed in this paper. In H.265/HEVC standard, video is encoded into multiple slices, and the slices are decoded independently. Inspired by the feature, each slice is individually encrypted using RC4 stream cipher. The pseudorandom binary sequence (PRBS) for one slice is related to encoding parameters and the SHA-256 hash value of the corresponding slice header, thus ensuring the real-time update of the PRBS and increasing the resistance to chosen-plaintext attack. A two-rounds shifting algorithm is designed to scramble non-zero coefficients of the transform units (TUs), and then motion vector difference (MVD) parameters, quantized transform coefficients (QTCs) and intra prediction modes (IPMs) are selected for encryption. The combination of coefficient scrambling and syntax elements encryption further improves the encryption performance. The proposed scheme is slice-synchronized. Hence, it decrypts normally in the case of packet loss, and supports online real-time interaction. Furthermore, the simulation and analysis results show that the proposed algorithm has format compatibility, high security and low time complexity, which is a promising tool for video cryptographic applications.
Purpose: 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.