Search Results (279)

Virtual-reality exposure therapy (VRET) is an emerging treatment for psychiatric disorders that enables immersive and controlled exposure to anxiety-provoking stimuli. Recent developments integrate real-time physiological monitoring, including heart rate (HR), electrodermal activity (EDA), and electroencephalography (EEG), to dynamically tailor therapeutic interventions. This systematic review examines studies that combine VRET with physiological data to adapt virtual environments in real time. A comprehensive search of major databases identified fifteen studies meeting the inclusion criteria: all employed physiological monitoring and adaptive features, with ten using biofeedback to modulate exposure based on single or multimodal physiological measures. The remaining studies leveraged physiological signals to inform scenario selection or threat modulation using dynamic categorization algorithms and machine learning. Although findings currently show an overrepresentation of anxiety disorders, recent studies are increasingly involving more diverse clinical populations. Results suggest that adaptive VRET is technically feasible and offers promising personalization benefits; however, the limited number of studies, methodological variability, and small sample sizes constrain broader conclusions. Future research should prioritize rigorous experimental designs, standardized outcome measures, and greater diversity in clinical populations. Adaptive VRET represents a frontier in precision psychiatry, where real-time biosensing and immersive technologies converge to enhance individualized mental health care. Full article

In the field of cognitive workload assessment for aerospace training, existing methods exhibit significant limitations in unimodal feature extraction and in leveraging complementary synergy among multimodal signals, while current fusion paradigms struggle to effectively capture nonlinear dynamic coupling characteristics across modalities. This study proposes DST-Net (Cross-Modal Downsampling Transformer Network), which synergistically integrates pilots’ multimodal physiological signals (electromyography, electrooculography, electrodermal activity) with flight dynamics data through an Anti-Aliasing and Average Pooling LSTM (AAL-LSTM) data fusion strategy combined with cross-modal attention mechanisms. Evaluation on the “CogPilot” dataset for flight task difficulty prediction demonstrates that AAL-LSTM achieves substantial performance improvements over existing approaches (AUC = 0.97, F1 Score = 94.55). Given the dataset’s frequent sensor data missingness, the study further enhances simulated flight experiments. By incorporating eye-tracking features via cross-modal attention mechanisms, the upgraded DST-Net framework achieves even higher performance (AUC = 0.998, F1 Score = 97.95) and reduces the root mean square error (RMSE) of cumulative flight error prediction to 1750. These advancements provide critical support for safety-critical aviation training systems. Full article

Traditional culture plays a vital role in shaping national identity and emotional belonging, making it imperative to explore innovative strategies for its digital preservation and engagement. This study investigates how interaction design in cultural digital games influences users’ emotional experiences and cultural understanding. Centering on the Chinese intangible cultural heritage puppet manipulation, we developed an interactive cultural game with three modes: gesture-based interaction via Leap Motion, keyboard control, and passive video viewing. A multimodal evaluation framework was employed, integrating subjective questionnaires with physiological indicators, including Functional Near-Infrared Spectroscopy (fNIRS), infrared thermography (IRT), and electrodermal activity (EDA), to assess users’ emotional responses, immersion, and perception of cultural content. Results demonstrated that gesture-based interaction, which aligns closely with the embodied cultural behavior of puppet manipulation, significantly enhanced users’ emotional engagement and cultural comprehension compared to the other two modes. Moreover, fNIRS data revealed broader activation in brain regions associated with emotion regulation and cognitive control during gesture interaction. These findings underscore the importance of culturally congruent interaction design in enhancing user experience and emotional resonance in digital cultural applications. This study provides empirical evidence supporting the integration of cultural context into interaction strategies, offering valuable insights for the development of emotionally immersive systems for intangible cultural heritage preservation. Full article

This study proposes a machine learning-based framework for classifying occupational stress levels among nurses using physiological time-series data collected from wearable sensors. The dataset comprises multimodal signals including electrodermal activity (EDA), heart rate (HR), skin temperature (TEMP), and tri-axial accelerometer measurements (X, Y, Z), which are labeled into three categorical stress levels: low (0), medium (1), and high (2). To enhance the usability of the raw data, a resampling process was performed to aggregate the measurements into one-minute intervals, followed by the application of the Synthetic Minority Over-sampling Technique (SMOTE) to mitigate severe class imbalance. Subsequently, a comparative classification analysis was conducted using four supervised learning algorithms: Random Forest, XGBoost, k-Nearest Neighbors (k-NN), and LightGBM. Model performances were evaluated based on accuracy, weighted F1-score, and confusion matrices to ensure robustness across imbalanced class distributions. Additionally, temporal pattern analyses by the day of the week and the hour of the day revealed significant trends in stress variation, underscoring the influence of circadian and organizational factors. Among the models tested, ensemble-based methods, particularly Random Forest and XGBoost with optimized hyperparameters, demonstrated a superior predictive performance. These findings highlight the feasibility of integrating real-time, sensor-driven stress monitoring systems into healthcare environments to support proactive workforce management and improve care quality. Full article

Ergonomic risk classification during manual lifting tasks is crucial for the prevention of workplace injuries. This study addresses the challenge of classifying lifting task risk levels (low, medium, and high risk, labeled as 0, 1, and 2) using multi-modal time-series data comprising raw facial landmarks and bio-signals (electrocardiography [ECG] and electrodermal activity [EDA]). Classifying such data presents inherent challenges due to multi-source information, temporal dynamics, and class imbalance. To overcome these challenges, this paper proposes a Multi-Adaptive Functional Neural Network (Multi-AdaFNN), a novel method that integrates functional data analysis with deep learning techniques. The proposed model introduces a novel adaptive basis layer composed of micro-networks tailored to each individual time-series feature, enabling end-to-end learning of discriminative temporal patterns directly from raw data. The Multi-AdaFNN approach was evaluated across five distinct dataset configurations: (1) facial landmarks only, (2) bio-signals only, (3) full fusion of all available features, (4) a reduced-dimensionality set of 12 selected facial landmark trajectories, and (5) the same reduced set combined with bio-signals. Performance was rigorously assessed using 100 independent stratified splits (70% training and 30% testing) and optimized via a weighted cross-entropy loss function to manage class imbalance effectively. The results demonstrated that the integrated approach, fusing facial landmarks and bio-signals, achieved the highest classification accuracy and robustness. Furthermore, the adaptive basis functions revealed specific phases within lifting tasks critical for risk prediction. These findings underscore the efficacy and transparency of the Multi-AdaFNN framework for multi-modal ergonomic risk assessment, highlighting its potential for real-time monitoring and proactive injury prevention in industrial environments. Full article

Decomposing Electrodermal Activity (EDA) into phasic (short-term, stimulus-linked responses) and tonic (longer-term baseline) components is essential for extracting meaningful emotional and physiological biomarkers. This study presents a comparative analysis of knowledge-driven, statistical, and deep learning-based methods for EDA signal decomposition, with a focus on in-the-wild data collected from wearable devices. In particular, the authors introduce the Feel Transformer, a novel Transformer-based model adapted from the Autoformer architecture, designed to separate phasic and tonic components without explicit supervision. The model leverages pooling and trend-removal mechanisms to enforce physiologically meaningful decompositions. Comparative experiments against methods such as Ledalab, cvxEDA, and conventional detrending show that the Feel Transformer achieves a balance between feature fidelity (SCR frequency, amplitude, and tonic slope) and robustness to noisy, real-world data. The model demonstrates potential for real-time biosignal analysis and future applications in stress prediction, digital mental health interventions, and physiological forecasting. Full article

This systematic review explores the role of smartwatches in stress management, mental health monitoring, and overall well-being. Drawing from 61 peer-reviewed studies published between 2016 and 2025, this review synthesizes empirical findings across diverse methodologies, including biometric data collection, machine learning algorithms, and user-centered design evaluations. Smartwatches, equipped with sensors for physiological signals such as heart rate, heart rate variability, electrodermal activity, and skin temperature, have demonstrated promise in detecting and predicting stress and mood fluctuations in both clinical and everyday contexts. This review emphasizes the need for interdisciplinary collaboration to advance technological precision, ethical data handling, and user experience design. Moreover, it highlights how different algorithms—such as Support Vector Machines (SVMs), Random Forests, Deep Neural Networks, and Boosting methods—perform across various physiological signals (e.g., HRV, EDA, skin temperature). Furthermore, it identifies performance trends and challenges across lab-based vs. real-world deployments, emphasizing the trade-off between generalizability and personalization in model design. Full article

Cockpit noise, as a critical environmental factor affecting flight safety, may impair pilots’ cognitive functions, leading to a decreased operational performance and decision-making errors, thereby posing potential threats to aviation safety. In order to reveal the relationship between the cockpit noise sound pressure level and pilot physiological indicators, and provide a scientific basis for cockpit noise airworthiness standards, this experiment takes pilot trainees as the research subject. Based on the principle of multimodal data synchronization, a sound field reconstruction system is used to reconstruct the cockpit sound field. Electroencephalogram (EEG), electrocardiogram (ECG), and electrodermal activity (EDA) measurements are carried out in different sound pressure level noise operating environments. The results show that with the increase in the sound pressure level, the significant suppression of α-wave activity in the occipital and parietal regions suggests that the cortical resting state is lifted and visual attention is enhanced; the enhancement of the β-wave in the frontal regions reflects the enhancement of alertness and prefrontal executive control, and the suppression of θ-wave activity in the frontal and temporal regions may indicate that cognitive tuning is suppressed, which reflects the brain’s rapid adaptive response to external noise stimuli in a high-noise environment; noise exposure triggers sustained sympathetic nerve hyperactivity, which is manifested by a significant acceleration of the heart rate and a significant increase in the mean value of skin conductance when the noise sound pressure level exceeds 70 dB(A). The correlation analysis between physiological indicators shows that cockpit noise has a multi-system synergistic effect on human physiological indicators. The experimental results indicate that noise has a significant impact on EEG, ECG, and EDA indicators. Full article

This study investigates the feasibility of using a wearable system with full-wave alternating current (AC) excitation to measure electrodermal activity (EDA). Typically measured using direct current (DC) excitation, EDA is often affected by signal drift due to electrode–skin polarisation. To address this, a portable device was developed that applies fixed-amplitude, full-wave AC signals and records EDA under controlled conditions. The electrical behaviour of the skin was also simulated using a multilayer model to analyse current propagation at different frequencies. The experimental procedure was conducted with ten healthy participants under controlled conditions. Two stages were carried out: the first compared the similarity of the skin conductance level (SCL) between DC and half-wave alternating current (AC) signals; the second analysed signal stability and skin response at full-wave AC excitation. Compared to DC, full-wave AC excitation demonstrated reduced signal drift, greater temporal stability, and enhanced measurement of the skin’s capacitive response. These findings support the adoption of AC excitation for EDA measurement, especially in ambulatory and real-time biomechanical applications where signal reliability and stability are essential. Full article

Background: Understanding the impact of different learning materials in terms of comprehension and engagement is essential for optimizing educational strategies. While digital learning tools are increasingly used, offering and multiplying different educational solutions, their effects on learners’ mental workload, attention, and engagement remain underexplored. This study aims to investigate how different types of learning content—educational videos, academic videos, and text reading—affect cognitive processing and engagement. Methods: Neurophysiological signals, including electroencephalography (EEG), electrodermal activity (EDA), and photoplethysmography (PPG), were recorded from experimental participants while they were engaged with each learning content. Subjective assessments of cognitive effort and engagement, together with a quiz to assess the knowledge acquisition, were collected through questionnaires for each tested content. Key neurophysiological metrics, such as engagement and Human Distraction Index (HDI), were computed and compared across conditions. Results: Our findings indicate that video-based learning materials, particularly educational videos with visual enhancements, elicited higher engagement and lower cognitive load compared to text-based learning. The text reading condition was associated with increased mental workload and a higher distraction index, suggesting greater cognitive demands. Correlation analyses confirmed strong associations between neurophysiological indicators and subjective evaluations. Conclusions: The results highlight the potential of neurophysiological measures to objectively assess learning experiences, paving the way for designing more effective and engaging learning platforms. Full article

As a forward-looking development in air traffic control (ATC), remote towers rely on virtualized information presentation, which may exacerbate visual fatigue among controllers and compromise operational safety. This study proposes a visual fatigue recognition model based on multimodal physiological signals. A 60-min simulated remote tower task was conducted with 36 participants, during which eye-tracking (ET), electroencephalography (EEG), electrocardiography (ECG), and electrodermal activity (EDA) signals were collected. Subjective fatigue questionnaires and objective ophthalmic measurements were also recorded before and after the task. Statistically significant features were identified through paired t-tests, and fatigue labels were constructed by combining subjective and objective indicators. LightGBM was then employed to rank feature importance by integrating split frequency and information gain into a composite score. The top 12 features were selected and used to train a multilayer perceptron (MLP) for classification. The model achieved an average balanced accuracy of 0.92 and an F1 score of 0.90 under 12-fold cross-validation, demonstrating excellent predictive performance. The high-ranking features spanned four modalities, revealing typical physiological patterns of visual fatigue across ocular behavior, cortical activity, autonomic regulation, and arousal level. These findings validate the effectiveness of multimodal fusion in modeling visual fatigue and provide theoretical and technical support for human factor monitoring and risk mitigation in remote tower environments. Full article

In recent years, the need to design inclusive workplaces has grown, particularly in manufacturing contexts where high cognitive demands may disadvantage neurodiverse individuals. In manufacturing environments, neurodiverse workers often experience difficulties processing standard instructions, increasing cognitive load and errors and reducing overall performance. This study proposes a methodology to assess cognitive load during assembly tasks to support workers with dyslexia. A multi-layer fuzzy logic framework was developed, integrating physiological, environmental, and task-related data. Physiological signals, including heart rate, heart rate variability, electrodermal activity, and eye-tracking data, were collected using wearable sensors. Ambient conditions were also measured. The model emphasizes the Reading dimension of cognitive load, critical for dyslexic individuals challenged by text-based instructions. A controlled laboratory study with 18 neurotypical participants simulated dyslexia scenarios with and without support, compared to a control condition. Results indicated that a lack of support increased cognitive load and reduced performance in complex tasks. In simpler tasks, control participants showed higher cognitive effort, possibly employing overcompensation strategies by exerting additional cognitive resources to maintain performance. Support mechanisms, such as audio prompts, effectively reduced cognitive load, highlighting the framework’s potential for fostering inclusive practices in industrial environments. Full article

This study investigates which statistical measures of electrodermal activity (EDA) signal features most effectively differentiate between responses to stimuli and resting states in participants performing tasks with varying cognitive and emotional reactions. The study involved 30 healthy participants. Collected EDA data were statistically analyzed, comparing the effectiveness of twelve statistical signal measures in detecting stimulus-induced changes. The aim of this study is to answer the following research question: Which statistical features of the electrodermal activity signal most effectively indicate changes induced by cognitive and emotional reactions, and are there such significant similarities (high correlations) among these features that some of them can be considered redundant? The results indicated that amplitude-related measures—mean, median, maximum, and minimum—were most effective. It was also found that some signal features were highly correlated, suggesting the possibility of simplifying the analysis by choosing just one measure from each correlated pair. The results indicate that stronger emotional stimuli lead to more pronounced changes in EDA than stimuli with a low emotional load. These findings may contribute to the standardization of EDA analysis in future research on cognitive and emotional reaction engagement. Full article

This study explores nursing students’ stress responses while they are being trained in a mixed reality (MR) setting that replicates highly stressful clinical scenarios. Using measurements of physiological indices such as heart rate, electrodermal activity, and skin temperature, the study assesses the level of stress when the students interact with digital patients whose vital signs and symptoms interact dynamically to respond to student inputs. The simulation consists of six segments, during which critical events like hypotension and hypoxia occur, and the patient’s condition changes based on the nurse’s clinical decisions. Machine learning algorithms were then used to analyze the nurse’s physiological data and to classify different levels of stress. Among the models tested, the Stacking Classifier demonstrated the highest classification accuracy of 96.4%, outperforming both Random Forest (96.18%) and Gradient Boosting (95.35%). The results showed clear patterns of stress during the simulation segments. Statistical analysis also found significant differences in stress responses and identified key physiological markers linked to each stress level. This pioneering study demonstrates the effectiveness of MR as a training tool for healthcare professionals in high-pressured scenarios and lays the groundwork for further studies on stress management, adaptive training procedures, and real-time detection and intervention in MR-based nursing training. Full article

Continuous monitoring of glucose levels is important for diabetes management and prevention. While traditional glucose monitoring methods are often invasive and expensive, recent approaches using machine learning (ML) models have explored non-invasive alternatives—but many still depend on manually logged food intake and activity, which is burdensome and impractical for everyday use. In this study, we propose a novel approach that eliminates the need for manual input by utilizing only passively collected, automatically recorded multi-modal data from non-invasive wearable sensors. This enables practical and continuous glucose prediction in real-world, free-living environments. We used the BIG IDEAs Lab Glycemic Variability and Wearable Device Data (BIGIDEAs) dataset, which includes approximately 26,000 CGM readings, simultaneous ly collected wearable data, and demographic information. A total of 236 features encompassing physiological, behavioral, circadian, and demographic factors were constructed. Feature selection was conducted using random-forest-based importance analysis to select the most relevant features for model training. We evaluated the effectiveness of various ML regression techniques, including linear regression, ridge regression, random forest regression, and XGBoost regression, in terms of prediction and clinical accuracy. Biological sex, circadian rhythm, behavioral features, and tonic features of electrodermal activity (EDA) emerged as key predictors of glucose levels. Tree-based models outperformed linear models in both prediction and clinical accuracy. The XGBoost (XR) model performed best, achieving an R-squared of 0.73, an RMSE of 11.9 mg/dL, an NRMSE of 0.52 mg/dL, a MARD of 7.1%, and 99.4% of predictions falling within Zones A and B of the Clarke Error Grid. This study demonstrates the potential of combining feature engineering and tree-based ML regression techniques for continuous glucose monitoring using wearable sensors. Full article