Search Results (377)

Accurate automatic sleep stage classification from single-channel EEG remains challenging due to the need for effective extraction of multiscale neurophysiological features and modeling of long-range temporal dependencies. This study aims to address these limitations by developing an efficient and compact deep learning architecture tailored for wearable and edge device applications. We propose MultiScaleSleepNet, a hybrid convolutional neural network–bidirectional long short-term memory–transformer architecture that extracts multiscale temporal and spectral features through parallel convolutional branches, followed by sequential modeling using a BiLSTM memory network and transformer-based attention mechanisms. The model obtained an accuracy, macro-averaged F1 score, and kappa coefficient of 88.6%, 0.833, and 0.84 on the Sleep-EDF dataset; 85.6%, 0.811, and 0.80 on the Sleep-EDF Expanded dataset; and 84.6%, 0.745, and 0.79 on the SHHS dataset. Ablation studies indicate that attention mechanisms and spectral fusion consistently improve performance, with the most notable gains observed for stages N1, N3, and rapid eye movement. MultiScaleSleepNet demonstrates competitive performance across multiple benchmark datasets while maintaining a compact size of 1.9 million parameters, suggesting robustness to variations in dataset size and class distribution. The study supports the feasibility of real-time, accurate sleep staging from single-channel EEG using parameter-efficient deep models suitable for portable systems. Full article

Sleep monitoring is critical for diagnosing and treating sleep disorders. Although polysomnography (PSG) remains the clinical gold standard, its complexity, discomfort, and lack of portability limit its applicability for long-term and home-based monitoring. To overcome these challenges, this study introduces a novel flexible headband system designed for multi-modal physiological signal acquisition, incorporating dry electrodes, a six-axis inertial measurement unit (IMU), and a temperature sensor. The device supports eight EEG channels and enables wireless data transmission via Bluetooth, ensuring user convenience and reliable long-term monitoring in home environments. To rigorously evaluate the system’s performance, we conducted comprehensive assessments involving 13 subjects over two consecutive nights, comparing its outputs with conventional PSG. Experimental results demonstrate the system’s low power consumption, ultra-low input noise, and robust signal fidelity, confirming its viability for overnight sleep tracking. Further validation was performed using the self-collected HBSleep dataset (over 184 h recordings of the 13 subjects), where state-of-the-art sleep staging models (DeepSleepNet, TinySleepNet, and AttnSleepNet) were applied. The system achieved an overall accuracy exceeding 75%, with AttnSleepNet emerging as the top-performing model, highlighting its compatibility with advanced machine learning frameworks. These results underscore the system’s potential as a reliable, comfortable, and practical solution for accurate sleep monitoring in non-clinical settings. Full article

This study proposes a Personalized Deep Learning-Based Sleep Recommendation System Using Lifelog Data (PDSRS-LD). Traditional sleep research primarily relies on bio signals such as EEG and ECG recorded during sleep but often fails to sufficiently reflect the influence of daily activities on sleep quality. To address this limitation, we collect lifelog data such as stress levels, fatigue, and sleep satisfaction via wearable devices and use them to construct individual user profiles. Subsequently, real sleep data obtained from an AI-powered motion bed are incorporated for secondary training to enhance recommendation performance. PDSRS-LD considers comprehensive user data, including gender, age, and physical activity, to analyze the relationships among sleep quality, stress, and fatigue. Based on this analysis, the system provides personalized sleep improvement strategies. Experimental results demonstrate that the proposed system outperforms existing models in terms of F1 score and Average Precision (mAP). These results suggest that PDSRS-LD is effective for real-time, user-centric sleep management and holds significant potential for integration into future smart healthcare systems. Full article

Background/Objectives: Sleep state misperception (SSM) is a common phenomenon in insomnia disorder (ID), characterized by a discrepancy between subjective and objective sleep metrics. Recent studies have revealed microstructural EEG alterations specifically in misperceiving ID patients, yet clinically accessible SSM markers remain limited. This study aimed to characterize SSM within ID by integrating standard polysomnography (PSG) features and cognitive-affective traits, focusing on accessible clinical tools. Methods: Twenty patients with ID and twenty healthy controls (HC) underwent a night of PSG recording and completed both sleep diaries and a comprehensive psychological assessment. SSM was quantified using the Total Sleep Time misperception index (TSTm), analyzed both dimensionally and categorically Results: IDs reported significantly altered sleep parameters compared to HCs, both subjectively and objectively. Within the ID sample, although underestimators and normoestimators had similar objective TST, underestimators showed significantly more cortical arousal density (CAd), a higher percentage of sleep stage 1 and higher non-acceptance of emotions. Notably, none of the HC reached the threshold for being classified as underestimators. Regression analyses identified CAd, latency to sleep stage 3 and to REM, percentage of REM sleep and lack of emotional clarity, as key predictors of TSTm. Conclusions: SSM in insomnia reflects a dimensional vulnerability grounded in subtle sleep fragmentation and emotional dysregulation. Recognizing SSM as a clinically meaningful phenomenon may guide more targeted, emotion-focused, interventions for insomnia. Full article

Advancements in electroencephalography (EEG) technology and feature extraction methods have paved the way for wearable, non-invasive systems that enable continuous sleep monitoring outside clinical environments. This study presents the development and evaluation of an EEG-based acquisition system for sleep staging, which can be adapted for wearable applications. The system utilizes a custom experimental setup with the ADS1299EEG-FE-PDK evaluation board to acquire EEG signals from the forehead and in-ear regions under various conditions, including visual and auditory stimuli. Afterward, the acquired signals were processed to extract a wide range of features in time, frequency, and non-linear domains, selected based on their physiological relevance to sleep stages and disorders. The feature set was reduced using the Minimum Redundancy Maximum Relevance (mRMR) algorithm and Principal Component Analysis (PCA), resulting in a compact and informative subset of principal components. Experiments were conducted on the Bitbrain Open Access Sleep (BOAS) dataset to validate the selected features and assess their robustness across subjects. The feature set extracted from a single EEG frontal derivation (F4-F3) was then used to train and test a two-step deep learning model that combines Long Short-Term Memory (LSTM) and dense layers for 5-class sleep stage classification, utilizing attention and augmentation mechanisms to mitigate the natural imbalance of the feature set. The results—overall accuracies of 93.5% and 94.7% using the reduced feature sets (94% and 98% cumulative explained variance, respectively) and 97.9% using the complete feature set—demonstrate the feasibility of obtaining a reliable classification using a single EEG derivation, mainly for unobtrusive, home-based sleep monitoring systems. Full article

Animal studies show that sleep regulation depends on subcortical networks, but whether the connectivity between subcortical areas contributes to human sleep variability remains unclear. We investigated whether the effective connectivity between the LC and hypothalamic subparts during wakefulness relates to sleep electrophysiology. Thirty-three younger (~22 y, 27 women) and 18 late middle-aged (~61 y, 14 women) healthy individuals underwent 7-Tesla functional MRI during wakefulness to assess LC–hypothalamus effective connectivity. Additionally, sleep EEG was recorded at night in the lab to examine the relationships between effective connectivity measures and REM sleep theta energy as well as sigma power prior to REM. Connectivity analyses revealed strong mutual positive influences between the LC and both the anterior–superior and posterior hypothalamus, consistent with animal studies. Aging was negatively associated with the connectivity from the anterior–superior hypothalamus (including the preoptic area) to the LC. In late middle-aged adults, but not younger adults, stronger effective connectivity from the anterior–superior hypothalamus to the LC was associated with lower REM theta energy. This association extended to other low-frequency bands during REM and NREM sleep. These findings highlight the age-dependent modulation of LC–hypothalamus interactions and their potential roles in sleep regulation, providing new insights into neural mechanisms underlying age-related sleep changes. Full article

With the rapid development of modern industry, people’s living pressures are gradually increasing, and an increasing number of individuals are affected by sleep disorders such as insomnia, hypersomnia, and sleep apnea syndrome. Many cardiovascular and psychiatric diseases are also closely related to sleep. Therefore, the early detection, accurate diagnosis, and treatment of sleep disorders an urgent research priority. Traditional manual sleep staging methods have many problems, such as being time-consuming and cumbersome, relying on expert experience, or being subjective. To address these issues, researchers have proposed multiple algorithmic strategies for sleep staging automation based on deep learning in recent years. This paper studies MASleepNet, a sleep staging neural network model that integrates multimodal deep features. This model takes multi-channel Polysomnography (PSG) signals (including EEG (Fpz-Cz, Pz-Oz), EOG, and EMG) as input and employs a multi-scale convolutional module to extract features at different time scales in parallel. It then adaptively weights and fuses the features from each modality using a channel-wise attention mechanism. The integrated temporal features are integrated into a Bidirectional Long Short-Term Memory (BiLSTM) sequence encoder, where an attention mechanism is introduced to identify key temporal segments. The final classification result is produced by the fully connected layer. The proposed model was experimentally evaluated on the Sleep-EDF dataset (consisting of two subsets, Sleep-EDF-78 and Sleep-EDF-20), achieving classification accuracies of 82.56% and 84.53% on the two subsets, respectively. These results demonstrate that deep models that integrate multimodal signals and an attention mechanism offer the possibility to enhance the efficiency of automatic sleep staging compared to cutting-edge methods. Full article

Analyzing electroencephalographic (EEG) time series can be challenging, especially with deep neural networks, due to the large variability among human subjects and often small datasets. To address these challenges, various strategies, such as self-supervised learning, have been suggested, but they typically rely on extensive empirical datasets. Inspired by recent advances in computer vision, we propose a pretraining task termed “frequency pretraining” to pretrain a neural network for sleep staging by predicting the frequency content of randomly generated synthetic time series. Our experiments demonstrate that our method surpasses fully supervised learning in scenarios with limited data and few subjects, and matches its performance in regimes with many subjects. Furthermore, our results underline the relevance of frequency information for sleep stage scoring, while also demonstrating that deep neural networks utilize information beyond frequencies to enhance sleep staging performance, which is consistent with previous research. We anticipate that our approach will be advantageous across a broad spectrum of applications where EEG data is limited or derived from a small number of subjects, including the domain of brain-computer interfaces. Full article

Background/Objective: Sleep is significant for human mental and physical health. Sleep disorders represent a crucial risk to human health, and a large portion of the world population suffers from them. The efficient identification of sleep disorders is significant for effective treatment. However, the precise and automatic detection of sleep disorders remains challenging due to the inter-subject variability, overlapping symptoms, and reliance on single-modality physiological signals. Methods: To address these challenges, a Neurophysiological Residual Gated Attention Multimodal Transformer Encoder (NRGAMTE) model was developed for robust sleep disorder detection using multimodal physiological signals, including Electroencephalogram (EEG), Electromyogram (EMG), and Electrooculogram (EOG). Initially, raw signals are segmented into 30-s windows and processed to capture the significant time- and frequency-domain features. Every modality is independently embedded by a One-Dimensional Convolutional Neural Network (1D-CNN), which preserves signal-specific characteristics. A Modality-wise Residual Gated Cross-Attention Fusion (MRGCAF) mechanism is introduced to select significant cross-modal interactions, while the learnable residual path ensures that the most relevant features are retained during the gating process. Results: The developed NRGAMTE model achieved an accuracy of 94.51% on the Sleep-EDF expanded dataset and 99.64% on the Cyclic Alternating Pattern (CAP Sleep database), significantly outperforming the existing single- and multimodal algorithms in terms of robustness and computational efficiency. Conclusions: The results shows that NRGAMTE obtains high performance across multiple datasets, significantly improving detection accuracy. This demonstrated their potential as a reliable tool for clinical sleep disorder detection. Full article

Sleep disorders are highly prevalent in children with Autism Spectrum Disorder (ASD), profoundly impacting their neurodevelopment and daily functioning. Alterations in sleep architecture and regulatory mechanisms contribute to difficulties with sleep onset, maintenance, and overall sleep quality. Sensory processing differences, commonly observed in ASD, may further exacerbate these disturbances by affecting arousal regulation and environmental responsiveness during sleep. Given the fundamental role of sleep in brain maturation, its disruption negatively impacts synaptic plasticity and neurological development, particularly during critical periods. These sleep-related alterations can influence cognitive and behavioral outcomes and may serve as early indicators of ASD, highlighting their potential value in early diagnosis and intervention. Understanding the neurobiological mechanisms linking sleep and ASD is essential for developing targeted therapeutic strategies. Ongoing research increasingly focuses on pharmacological, nutraceutical, and behavioral interventions aimed at mitigating sleep disorders and their cascading effects on neurodevelopment. Optimizing these therapeutic approaches through a multidisciplinary lens is crucial for enhancing clinical outcomes and improving overall quality of life in children with ASD. Full article

Background/Objectives: Modern technology is transforming the field of massage, enhancing relaxation and wellness through innovative devices. The aim of the present study was to examine the effect of various massage protocols available using an automatic electric massage chair (AEMC) prior to daytime napping on relaxation and indices of sleep quality. Methods: This study is a randomized, single-blind, placebo-controlled, four arm, interventional clinical trial. A total of 12 healthy individuals (21.8 ± 2.2 years, 6 F/6 M) were randomly assigned to four different groups: (1) the control (CON) session involving a 30 min rest on an automatic switch-off massage chair, (2) the easy-sleep (ES) massage session designed to promote sleep, (3) the fatigue-recovery (FR) massage session designed to reduce muscle fatigue, and (4) the worker-mode (WM) massage session designed to promote muscle relaxation. During the four sessions, participants sat in the massage chair for 30 min, followed immediately by an additional 30 min period of lying down on a standard double bed. Brain activity was monitored using a polysomnography EEG system, while validated tests and questionnaires assessed vitals and the state of relaxation. Results: The ES massage significantly reduced muscle tone by 12% and heart rate by 22% (p = 0.008 and p = 0.007, respectively). Additionally, it increased subjective sleepiness by 4.5% and sleep efficiency by 5.7% compared to the results for the control condition (p ≤ 0.005). Conclusions: It is evident that the use of an AEMC can reduce tension and improve feelings of relaxation. The easy-sleep program seems to be a promising non-pharmacological approach for enhancing relaxation and promoting daytime sleep, acting as a non-pharmacological tool to reduce stress, improve sleep quality, and promote workplace well-being. The trial was registered as NCT06784700. Full article

Sleepiness at the wheel is an important contributor to road traffic accidents. Slow eye movement (SEM) serves as a reliable physiological indicator for the sleep onset period (SOP). To detect SEM for recognizing drivers’ SOP, a Parallel Multimodal CNN-Transformer (PMMCT) model is proposed. The model employs two parallel feature extraction modules to process bimodal signals, each comprising convolutional layers and Transformer encoder layers. The extracted features are fused and then classified using fully connected layers. The model is evaluated on two bimodal signal combinations HEOG + O2 and HEOG + HSUM, where HSUM is the sum of two single-channel horizontal electrooculogram (HEOG) signals and captures electroencephalograph (EEG) features similar to those in the conventional O2 channel. Experimental results indicate that using the PMMCT model, the HEOG + HSUM combination performs comparably to the HEOG + O2 combination and outperforms unimodal HEOG by 2.73% in F1-score, with average classification accuracy and F1-score of 99.89% and 99.35%, outperforming CNN, CNN-LSTM, and CNN-LSTM-Attention models. The model exhibits minimal false positives and false negatives, with average values of 5.2 and 0.8. By combining CNNs’ local feature extraction with Transformers’ global temporal modeling, and using only two HEOG electrodes, the system offers superior performance while enhancing wearable device comfort for real-world applications. Full article

The increasing demand for portable health monitoring has highlighted the need for automated sleep staging systems that are both accurate and computationally efficient. However, most existing deep learning models for electroencephalogram (EEG)-based sleep staging suffer from parameter redundancy, fixed dilation rates, and limited generalization, restricting their applicability in real-time and resource-constrained scenarios. In this paper, we propose ADG-SleepNet, a novel lightweight symmetry-aware multi-scale dilation-gated temporal convolutional network enhanced with adaptive attention mechanisms for EEG-based sleep staging. ADG-SleepNet features a structurally symmetric, parallel multi-branch architecture utilizing various dilation rates to comprehensively capture multi-scale temporal patterns in EEG signals. The integration of adaptive gating and channel attention mechanisms enables the network to dynamically adjust the contribution of each branch based on input characteristics, effectively breaking architectural symmetry when necessary to prioritize the most discriminative features. Experimental results on the Sleep-EDF-20 and Sleep-EDF-78 datasets demonstrate that ADG-SleepNet achieves accuracy rates of 87.1% and 85.1%, and macro F1 scores of 84.0% and 81.1%, respectively, outperforming several state-of-the-art lightweight models. These findings highlight the strong generalization ability and practical potential of ADG-SleepNet for EEG-based health monitoring applications. Full article

In addition to characteristic and easily identifiable behavioural signs—namely epileptic seizures—electroencephalography (EEG) has long been a standard component of epilepsy diagnosis protocols. In veterinary practice, EEG is typically performed in a semi-invasive manner, using subcutaneous electrodes and sedation. Here, we propose that the non-invasive polysomnography protocol, originally developed for basic research, can serve as a more welfare-friendly yet informative alternative for assessing epileptic brain activity in dogs. In this study, N = 11 family dogs diagnosed with epilepsy underwent a single non-invasive polysomnography session. EEG-based evidence for epileptic activity was detected in two cases. Polysomnography data from these 11 epileptic dogs were further analysed to evaluate sleep structure parameters. Compared to a matched control group of N = 11 clinically healthy dogs, the epileptic group exhibited reduced sleep efficiency, increased sleep latency, more wakings after sleep onset, and less time spent in drowsiness and non-REM sleep. These findings support the potential utility of non-invasive brain monitoring techniques, such as polysomnography, in the diagnosis and management of epilepsy in veterinary medicine. Full article

The automatic classification of sleep stages and Cyclic Alternating Pattern (CAP) subtypes from electroencephalogram (EEG) recordings remains a significant challenge in computational sleep research because of the short duration of CAP events and the inherent class imbalance in clinical datasets. Background/Objectives: The research introduces a domain-specific deep learning system that employs an LSTM network optimized through a PSO-Hyperband hybrid hyperparameter tuning method. Methods: The research enhances EEG-based sleep analysis through the implementation of hybrid optimization methods within an LSTM architecture that addresses CAP sequence classification requirements without requiring architectural changes. Results: The developed model demonstrates strong performance on the CAP Sleep Database by achieving 97% accuracy for REM and 96% accuracy for stage S0 and ROC AUC scores exceeding 0.92 across challenging CAP subtypes (A1–A3). The model transparency is improved through the application of SHAP-based interpretability techniques, which highlight the role of spectral and morphological EEG features in classification outcomes. Conclusions: The proposed framework demonstrates resistance to class imbalance and better discrimination between visually similar CAP subtypes. The results demonstrate how hybrid optimization methods improve the performance, generalizability, and interpretability of deep learning models for EEG-based sleep microstructure analysis. Full article