Search Results (178)

Accurate automatic sleep staging remains challenging due to complex temporal dynamics, inter-subject variability, and the difficulty of effectively integrating heterogeneous physiological signals. Electroencephalogram (EEG) and electrooculogram (EOG) recordings provide complementary information for sleep analysis; however, most existing multimodal approaches rely on simple feature concatenation, which limits their ability to capture structured inter-modality relationships. This paper proposes MFSleepNet, a multimodal sleep staging framework that explicitly models interactions between EEG and EOG signals. The proposed system incorporates a multimodal feature fusion module to enable bidirectional information exchange between modality-specific representations, followed by a gated temporal-channel attention mechanism to adaptively emphasize informative temporal segments and signal channels, facilitating joint representation learning while preserving modality-specific characteristics. Experiments on three public datasets (Sleep-EDF, SHHS, and HSP) under an epoch-level cross-validation protocol show that MFSleepNet consistently outperforms representative single-modality and multimodal baseline methods in terms of overall accuracy, Cohen’s κ, and Macro-F1. Ablation studies further demonstrate the contribution of each functional module. Correlation analysis indicates stage-dependent variations in EEG–EOG relationships, while interaction-based experiments show that explicit feature interaction improves both joint and modality-specific representations. Grad-CAM visualizations provide interpretability of model decisions. External validation on unseen subjects reveals a noticeable performance drop, highlighting the challenges of inter-subject variability and the limited baseline generalization capability of the model. To address this, a lightweight subject-specific adaptation strategy is introduced, which improves performance using a small amount of labeled subject-specific data. Overall, the proposed framework provides an effective and interpretable solution for multimodal sleep staging while emphasizing the importance of structured inter-modality interaction and subject-adaptive modeling in practical applications. Full article

This study quantitatively evaluates the performance of a non-invasive hybrid brain–computer interface (BCI) compared to a conventional mouse in pointing (point-and-click) tasks. A commercial wearable BCI (Brainfingers), based on electromyography (EMG) and electrooculography (EOG) signals with low-level electroencephalography (EEG) components, was assessed against a Microsoft Optical Mouse using ISO/TS 9241-411-based one-dimensional (1D) and two-dimensional (2D) target acquisition tasks. Pointer coordinates were recorded and analyzed using Fitts’ law metrics. A total of 48 non-disabled participants completed the experiments. The results reveal significant performance differences between the two input devices. The BCI device exhibits substantially lower performance than the mouse across the reported Fitts’ law measures. Mean throughput was 0.35 bits/s for the BCI and 6.03 bits/s for the mouse in the 1D tests and 0.43 bits/s for the BCI and 5.17 bits/s for the mouse in the 2D tests. Despite the BCI’s low performance and although the present experiments involved non-disabled participants, the findings, considered alongside the prior literature on Brainfingers and non-invasive BCIs for computer access, suggest that the device may still have assistive technology value for users with severe motor impairments. Full article

Objective: Single-channel EEG artifact suppression often relies on signal decomposition; however, it is not always clear how much clean EEG is recoverable from a given decomposition when component weighting is ideal. We present an oracle-based benchmark that characterises this best-case recoverability across common 1-D decomposition families under controlled EOG, EMG, and mixed contamination. This work does not propose a new denoising algorithm; rather, it isolates representation capacity from component-selection heuristics by computing an upper bound on reconstruction quality. Approach: Using EEGdenoiseNet, we constructed a synthetic benchmark of 4500 single-channel 2 s segments (125 Hz; T = 250) by mixing clean EEG with ocular (EOG) and/or cranial EMG exemplars at noise-to-signal ratios (NSRs) spanning −10 to +10 dB (floor −10 dB denotes an absent modality). We evaluated variational mode decomposition (VMD), singular spectrum analysis (SSA), discrete wavelet transform (DWT), and CEEMDAN by decomposing each mixture and reconstructing the clean EEG using a bounded nonnegative linear combination of components obtained via constrained least squares (the oracle). Main results: Under this oracle benchmark, SSA achieved the lowest reconstruction error in most tested conditions, while DWT tended to rank best in milder ocular regimes; VMD performance improved, with an increased mode count at higher computational cost. CEEMDAN exhibited higher latency dominated by ensemble settings. Significance: These results should be interpreted as decomposition-level upper bounds under controlled mixtures, not field-ready denoising performance. The benchmark provides a tool with which to compare representational recoverability across decompositions and to inform the subsequent design of practical component-selection strategies. Full article

Locked-in Syndrome (LIS) is a neurological condition in which individuals remain conscious but experience complete paralysis of voluntary muscles, except for eye movements—highlighting the need for reliable assistive communication technologies. This study presents the design and evaluation of an Arabic electrooculogram (EOG)-based communication system with adaptive classification capabilities for LIS applications. A custom-designed EOG acquisition circuit incorporating filtering and amplification stages was implemented and compared with the OpenBCI Cyton board. The system employed a hybrid classification approach combining amplitude, temporal, and statistical features to distinguish between blinks and voluntary vertical eye movements. Testing with ten healthy subjects yielded a mean classification accuracy of 83.96% ± 4.59% and an information transfer rate of 10.43 letters per minute, corresponding to a 30.38% improvement over conventional approaches. The custom-designed circuit achieved a signal-to-noise ratio of 25.21 dB, outperforming the OpenBCI Cyton board by 8% while reducing system cost by 62%. The integration with a Morse code-based interface enabled Arabic letter composition, while the system incorporated auto-completion and text-to-speech functionalities to further enhance communication efficiency. This cost-effective solution addresses a critical gap in assistive technologies for Arabic-speaking individuals with LIS and shows strong potential for enhancing their communication abilities and overall quality of life. Full article

Emotions and affective responses are core intervention targets in music therapy. Through acoustic elements, music can evoke emotional responses at physiological and neurological levels, influencing cognition and behavior while providing an important dimension for evaluating therapeutic efficacy. However, emotions are inherently abstract and difficult to represent directly. Artificial intelligence models therefore provide a promising tool for modeling and quantifying such abstract affective states from physiological signals. In this paper, we propose a structured and explicitly factorized multi-modal representation learning framework for joint affective state and preference inference. Instead of entangling heterogeneous dynamics within monolithic encoders, the framework decomposes representation learning into cross-channel interaction modeling and intra-channel temporal–spectral organization modeling. The framework integrates electroencephalography (EEG), peripheral physiological signals (GSR, BVP, EMG, respiration, and temperature), and eye-movement data (EOG) within a unified temporal modeling paradigm. At its core, a Dynamic Token Feature Extractor (DTFE) transforms raw time series into compact token representations and explicitly factorizes representation learning into (i) explicit channel-wise cross-series interaction modeling and (ii) temporal–spectral refinement via learnable frequency-domain gating. These complementary structural modules are implemented through Cross-Series Intersection (CSI) and Intra-Series Intersection (ISI), which perform low-rank channel dependency learning and adaptive spectral modulation, respectively. A hierarchical cross-modal fusion strategy integrates modality-level tokens in a representation-consistent and interaction-aware manner, enabling coordinated modeling of neural, autonomic, and attentional responses. The entire framework is optimized under a unified multi-task objective for valence, arousal, and liking prediction. Experiments on the DEAP dataset demonstrate consistent improvements over state-of-the-art methods. The model achieves 98.32% and 98.45% accuracy for valence and arousal prediction, 97.96% for quadrant classification in single-task evaluation, and 92.8%, 91.8%, and 93.6% accuracy for valence, arousal, and liking in joint multi-task settings. Overall, this work establishes a structure-aware and factorized multi-modal representation learning framework for robust affective decoding and intelligent music therapy systems. Full article

Rapid eye movement (REM) sleep is increasingly understood as a heterogeneous state composed of two neurophysiologically distinct microstates: tonic REM and phasic REM. Phasic REM, defined by brief clusters of saccadic eye movements and transient cortical activation, has been linked to emotional memory consolidation, sensorimotor integration, and autonomic modulation. Despite its importance, automated quantification of phasic versus tonic REM remains uncommon, mainly because existing electrooculography (EOG) methods rely on fixed thresholds or generic wavelet families that do not accurately capture real saccade morphology in clinical polysomnography (PSG). This study introduces a fully automated framework for detecting phasic REM based on hybrid adaptive segmentation of a single EOG channel. The segmentation algorithm fuses median absolute deviation (MAD) amplitude-change detection with a morphology score derived from a custom saccade kernel built from manually verified EyeCon recordings. Segment boundaries are refined using local derivative extrema to improve temporal alignment. A supervised support vector machine (SVM) classifier further refines segment labels using features based on saccade morphology, including correlations with custom log-sigmoid templates and a morphology similarity measure. All segmentation and classification hyperparameters were optimized exclusively on controlled EyeCon datasets with precise ground-truth event markers. The final model was then applied without modification to 21 full-night clinical PSG recordings. Event-level analysis on EyeCon yielded 92.9% correct detections, with 5.3% fragmentation and 1.8% missed events. When aggregated into saccadic bursts, the resulting REM microstructure was physiologically consistent: phasic REM accounted for 31.8 ± 3.5% of REM duration, and tonic REM for 68.2 ± 3.5%. Additional EEG analysis confirmed increased beta and gamma power during phasic REM, supporting physiological validity. The proposed framework provides an interpretable, morphology-aware, and computationally efficient tool for large-scale REM microstructure research. Its single-channel design and external validation on clinical PSG recordings make it suitable for both retrospective analyses and future clinical applications. Full article

Automatic sleep staging from polysomnography is challenged by marked spectro-temporal heterogeneity and non-stationary cross-channel artifacts, which often undermine naïve multimodal fusion. To address this, a Gated Cross-Modal and Temporal Fusion Network (G-CMTF Net) is proposed as an end-to-end model operating on 30 s EEG epochs and auxiliary EOG and EMG signals, in which cross-modal contributions are regulated through reliability-aware gating. A spectro-temporal disentanglement frontend learns multi-scale temporal features while incorporating FFT-derived band-power embeddings to preserve physiologically meaningful oscillatory cues. At the epoch level, gated fusion suppresses artifact-prone auxiliary inputs, thereby limiting noise transfer into a shared latent space. Long-range sleep dynamics are modeled via a convolution-augmented self-attention encoder that captures both local morphology and transition structure. On Sleep-EDF-20 and Sleep-EDF-78, G-CMTF Net achieves Macro-F1/ACC of 81.3%/85.5% and 78.2%/83.4%, respectively, while maintaining high sensitivity and geometric-mean performance on transitional epochs, consistent with the function of reliability-aware gated fusion under non-stationary auxiliary artifacts. From a symmetry perspective, the proposed framework enforces a structured balance between heterogeneous modalities by promoting representational consistency while adaptively suppressing asymmetric noise contributions. Full article

Interest in wind energy systems of different power ratings has increased significantly in recent years; however, low-power wind turbines are particularly sensitive to wind gust disturbances, which strongly affect their power electronic systems. In this work, a control strategy is proposed for regulating the output voltage of a buck converter integrated into a small wind turbine. To this end, a symmetric variable gain is incorporated into the classical sliding mode control framework, enabling the controller to dynamically adjust the control effort according to the operating conditions. The main objective of the proposed approach is to mitigate output voltage fluctuations induced by Extreme Operating Gusts (EOGs), which have a more pronounced impact on low-power wind turbines. The effectiveness of the proposed controller is validated through both simulation and experimental results. Full article

Steady-state visual evoked potentials (SSVEPs) are one of the key paradigms used in brain–computer interface (BCI) systems. Their performance, however, is substantially degraded by EEG artifacts of muscular, motion-related, and ocular origin. This issue is particularly pronounced in individuals exhibiting increased facial muscle tension or involuntary eye movements. The aim of this study was to develop and evaluate an EEG artifact reduction method based on auxiliary channels, including central (Cz), frontal (Fp1), electrooculographic (HEOG), and muscular electrodes (neck, cheek, jaw). Signals from these channels were used to model the physical sources of interference recorded concurrently with occipital brain activity (O1, O2, Oz). EEG signal cleaning was performed using linear regression in 1-s windows, followed by frequency-domain analysis to extract features related to stimulation frequencies and SSVEP classification using SVM and CNN algorithms. The experiment involved three visual stimulation frequencies (7, 8, and 9 Hz) generated by LEDs and the recording of controlled facial and jaw-related artifacts. Experiments conducted on 12 participants demonstrated a 9% increase in classification accuracy after artifact removal. Further analysis indicated that the Cz and jaw channels contributed most significantly to effective artifact suppression. The results confirm that the use of auxiliary channels substantially improves EEG signal quality and enhances the reliability of BCI systems under real-world conditions. Full article

Background: Amyotrophic lateral sclerosis (ALS) progressively impairs motor function, compromising speech and limiting communication. Augmentative and alternative communication (AAC) is essential to maintain autonomy, social participation, and quality of life for people with ALS (PALS). This review maps technological developments in AAC, from low-tech tools to advanced brain–computer interface (BCI) systems. Methods: We conducted a scoping review following the PRISMA extension for scoping reviews. PubMed, Web of Science, SciELO, MEDLINE, and CINAHL were screened for studies published up to 31 August 2025. Peer-reviewed RCT, cohort, cross-sectional, and conference papers were included. Single-case studies of invasive BCI technology for ALS were also considered. Methodological quality was evaluated using JBI Critical Appraisal Tools. Results: Thirty-seven studies met inclusion criteria. High-tech AAC—particularly eye-tracking systems and non-invasive BCIs—were most frequently studied. Eye tracking showed high usability but was limited by fatigue, calibration demands, and ocular impairments. EMG- and EOG-based systems demonstrated promising accuracy and resilience to environmental factors, though evidence remains limited. Invasive BCIs showed the highest performance in late-stage ALS and locked-in syndrome, but with small samples and uncertain long-term feasibility. No studies focused exclusively on low-tech AAC interventions. Conclusions: AAC technologies, especially BCIs, EMG and eye-tracking systems, show promise in supporting autonomy in PALS. Implementation gaps persist, including limited attention to caregiver burden, healthcare provider training, and the real-world use of low-tech and hybrid AAC. Further research is needed to ensure that communication solutions are timely, accessible, and effective, and that they are tailored to functional status, daily needs, social participation, and interaction with the environment. Full article

Bio-electric fields—manifested as Electroencephalogram (EEG), Electrocardiogram (ECG), Electromyogram (EMG), and Electrooculogram (EOG)—are fundamental to modern medical diagnostics but often suffer from severe data imbalance, scarcity, and environmental noise. Generative Adversarial Networks (GANs) offer a powerful, nonlinear solution to these modeling hurdles. This review presents a comprehensive survey of GAN methodologies specifically tailored for bio-electric signal processing. We first establish a theoretical foundation by detailing GAN principles, training mechanisms, and critical structural variants, including advancements in loss functions and conditional architectures. Subsequently, the paper extensively analyzes applications ranging from high-fidelity signal synthesis and noise reduction to multi-class classification. Special attention is given to clinical anomaly detection, specifically covering epilepsy, arrhythmia, depression, and sleep apnea. Furthermore, we explore emerging applications such as modal transformation, Brain–Computer Interfaces (BCI), de-identification for privacy, and signal reconstruction. Finally, we critically evaluate the computational trade-offs and stability issues inherent in current models. The study concludes by delineating prospective research avenues, emphasizing the necessity of interdisciplinary synergy to advance personalized medicine and intelligent diagnostic systems. Full article

Background and Objective: The accurate detection of sleep stages and disorders in older adults is essential for the effective diagnosis and treatment of sleep disorders affecting millions worldwide. Although Polysomnography (PSG) remains the primary method for monitoring sleep in medical settings, it is costly and time-consuming. Recent automated models have not fully explored and effectively fused the sleep features that are essential to identify sleep stages and disorders. This study proposes a novel automated model for detecting sleep stages and disorders in older adults by analyzing PSG recordings. PSG data include multiple channels, and the use of our proposed advanced methods reveals the potential correlations and complementary features across EEG, EOG, and EMG signals. Methods: In this study, we employed three novel advanced architectures, (1) CNNs, (2) CNNs with Bi-LSTM, and (3) CNNs with a transformer encoder, for the automatic classification of sleep stages and disorders using multichannel PSG data. The CNN extracts local features from RGB spectrogram images of EEG, EOG, and EMG signals individually, followed by an appropriate column-wise feature fusion block. The Bi-LSTM and transformer encoder are then used to learn and capture intra-epoch feature transition rules and dependencies. A residual connection is also applied to preserve the characteristics of the original joint feature maps and prevent gradient vanishing. Results: The experimental results in the CAP sleep database demonstrated that our proposed CNN with transformer encoder method outperformed standalone CNN, CNN with Bi-LSTM, and other advanced state-of-the-art methods in sleep stages and disorders classification. It achieves an accuracy of 95.2%, Cohen’s kappa of 93.6%, MF1 of 91.3%, and MGm of 95% for sleep staging, and an accuracy of 99.3%, Cohen’s kappa of 99.1%, MF1 of 99.2%, and MGm of 99.6% for disorder detection. Our model also achieves superior performance to other state-of-the-art approaches in the classification of N1, a stage known for its classification difficulty. Conclusions: To the best of our knowledge, we are the first group going beyond the standard to investigate and innovate a model architecture which is accurate and robust for classifying sleep stages and disorders in the elderly for both patient and non-patient subjects. Given its high performance, our method has the potential to be integrated and deployed into clinical routine care settings. Full article

REM sleep behavioral disorder (RBD) is of increasing interest in Parkinson’s disease (PD). Previous studies exploring the association between REM sleep without atonia (RWA) and clinical PD features or other objective sleep metrics are scarce and have used PSG findings. A mobile electroencephalography (EEG)/electrooculography (EOG) recording system with two channels can objectively measure sleep parameters, including RWA, during natural sleep at home. We investigated whether RWA measured on a portable recording device at home could be associated with clinical PD features or other sleep metrics using baseline data from the ZEAL study. Differences between patients with and without RWA was analyzed using ANCOVA test. REM sleep length was significantly longer in patients with RWA than in those without RWA. A multivariate comparison using ANCOVA showed a significant difference in log-transformed REM sleep duration of patients with RWA after adjustment for potential confounders (adjusted mean difference of 1.203; 95% confidence interval 0.468 to 1.937; p = 0.003). The strength of this study was that it evaluated the association between RWA during natural sleep at home and clinical variables as well as other sleep metrics. The major result was that patients with and without RWA did not differ in their clinical variables, and there was no relation between RWA and objective sleep metrics other than REM sleep. The duration of REM sleep may be associated with RWA during natural sleep at home. Full article

Electroencephalography (EEG) signal acquisition is often affected by artifacts, challenging applications such as brain disease diagnosis and Brain-Computer Interfaces (BCIs). This paper proposes TF-Denoiser, a deep learning model using a joint time-frequency optimisation strategy for artifact removal. The proposed method first employs a position embedding module to process EEG data, enhancing temporal feature representation. Then, the EEG signals are transformed from the time domain to the complex frequency domain via Fourier transform, and the real and imaginary parts are denoised separately. The multi-attention denoising module (MA-denoise) is used to extract both local and global features of EEG signals. Finally, joint optimisation of time-frequency features is performed to improve artifact removal performance. Experimental results demonstrate that TF-Denoiser outperforms the compared methods in terms of correlation coefficient (CC), relative root mean square error (RRMSE), and signal-to-noise ratio (SNR) on electromyography (EMG) and electrooculography (EOG) datasets. It effectively reduces ocular and muscular artifacts and improves EEG denoising robustness and system stability. Full article

STEM (Science, Technology, Engineering, and Mathematics) is essential for the development of 21st-century skills, particularly in a world driven by technological innovation. However, in vulnerable school contexts, access to meaningful STEM experiences remains limited. This study addresses this issue through the design and implementation of a didactic strategy in a public high school in Bogotá, Colombia, based on two educational resources: the BioSen electronic board, which is compatible with Arduino technology and designed to acquire physiological signals such as electrocardiography (ECG), electromyography (EMG), electrooculography (EOG), and body temperature; and the Space Exploration instructional guide, which is organized around contextualized learning missions. This study employed a quasi-experimental mixed-methods design that combined pre–post perception questionnaires, unstructured classroom observations, and a contextualized knowledge test administered to three student groups. Findings demonstrate that after eight weeks of implementation with upper secondary students, the strategy had a positive impact on the development of 21st-century skills, such as creativity, computational thinking, and critical thinking. These skills were assessed through a mixed quasi-experimental design combining perception questionnaires, qualitative observations, and knowledge evaluations. Unlike the control groups, students who participated in the intervention adjusted their self-perceptions when facing real-world challenges and showed progress in the application of key competencies. Overall, the results support the effectiveness of integrating low-cost biomedical tools with gamified STEM instruction to enhance higher-order thinking skills and student engagement in vulnerable educational contexts. Full article