Decoding EEG-based cognitive load using fusion of temporal and functional connectivity features

Alcohol consumption tends to increase during the transition from childhood to adolescence, and risk factors at the individual, family, or environmental level, as well as altered brain structure and function have been associated with the likelihood of developing alcohol use disorder (AUD) or binge drinking later in life. Most studies have focused on limited subsets of multilevel or neuroimaging factors, typically emphasizing alcohol initiation, binge drinking, or AUD in cross-sectional designs rather than exploring longitudinal alcohol consumption trajectories. Our study addresses these gaps by examining a comprehensive set of multilevel, multidomain factors and multimodal brain imaging features to prospectively predict early alcohol sipping trajectories over time with large data from the Adolescent Brain Cognitive Development (ABCD) study (baseline N = 11,686, mean age = 9.92 years). We applied machine learning methods to baseline individual, family, and environmental factors as well as structural and functional brain connectivity features, analyzing these predictors separately and in combination. Key findings reveal that functional connectivity features and multilevel factors distinguish individuals with an increasing alcohol sipping trajectory from those who initially experimented with alcohol but reduced their consumption over time. Moreover, we found important structural and functional features that predicted those who increasingly sipped over time versus the ones who did not engage in alcohol experimentation. Stable interactions between age and cortical thickness, surface area and functional connectivity between the amygdala and executive or default networks suggest that accelerated cortical maturation could predict a pattern of increasing alcohol sipping over time. These trends could inform us how individual, family, and environmental factors together with brain imaging features impact the development of different alcohol sipping trajectories over time.

Accurate demand forecasting is a critical enabler of sustainable Supply Chain Management, as it directly influences production planning, inventory control, resource utilization, and environmental impact. In alignment with the United Nations Sustainable Development Goals, particularly SDG 9 (Industry, Innovation and Infrastructure), SDG 12 (Responsible Consumption and Production), and SDG 13 (Climate Action), this study proposes a layered AI-based forecasting framework aimed at enhancing supply chain efficiency while supporting sustainable development objectives. The proposed framework integrates three progressively advanced models: a standard Long Short-Term Memory (LSTM) network, an LSTM with an Attention Mechanism, and a Quantum-inspired Attention LSTM, designed to capture long-term temporal dependencies and complex feature interactions inherent in real-world supply chain data. The framework is experimentally validated using three real-world supply chain datasets representing e-commerce, grocery retail, and customer analytics domains. Experimental results demonstrate that the Quantum-inspired Attention LSTM consistently outperforms classical LSTM and attention-based LSTM models, achieving forecasting accuracies of 92.85%, 92.56%, and 90.42%, respectively. Improved forecasting accuracy directly contributes to sustainability by reducing overproduction, minimizing excess inventory, lowering avoidable transportation and storage operations, and mitigating carbon emissions across supply chain networks. An ablation study further confirms that the incorporation of attention and quantum-inspired representations significantly enhances model robustness and predictive reliability. The findings illustrate that advanced AI-driven forecasting, when explicitly aligned with SDG-oriented objectives, can serve as a practical and scalable mechanism for enabling sustainable supply chain operations, supporting data-driven decision-making, and advancing responsible production and consumption practices.

Generalized anxiety disorder (GAD) is one of common anxiety disorders in adolescents. Although adolescents with GAD are thought to be at high risk for other mental diseases, the disease-specific alterations have not been adequately explored. Recent studies have revealed the abnormal functional connectivity (FC) in adolescents with GAD. Most previous researches have investigated the static FC which ignores the fluctuations of FC over time and focused on the structures of “fear circuit”. To figure out the alterations of dynamic FC caused by GAD and the possibilities of dynamic FC as biomarkers, we propose an effective approach to identify adolescent GAD using temporal features derived from dynamic FC. In our study, the instantaneous synchronization of pairwise signals was estimated as dynamic FC. The Hurst exponent (H) and variance, indicating regularity and variable degree of a time series respectively, were calculated as temporal features of dynamic FC. By leave-one-out cross-validation (LOOCV), a relatively high accuracy of 88.46% could be achieved when H and variance of dynamic FC were combined as features. In addition, we identified the disease-related regions, including regions belonging to default mode (DM) and cerebellar networks. The results suggest that temporal features of dynamic FC could achieve a clinically acceptable diagnostic power and serve as biomarkers of adolescent GAD. Furthermore, our work could be helpful in understanding the pathophysiological mechanism of adolescent GAD.

Brain-computer interface (BCI) systems create a novel communication channel from the brain to an output device bypassing conventional motor output pathways of nerves and muscles. Modern BCI technology is essentially based on techniques for the classification of single-trial brain signals. With respect to the topographic patterns of brainrhythm modulations, the common spatial patterns (CSPs) algorithm has been proven to be very useful to producesubject-specific and discriminative spatial filters; but it didn't consider temporal structures of event-related potentials which may be very important for single-trial EEG classification. In this paper, we propose a new framework offeature extraction for classification of hand movement imagery EEG. Computer simulations on real experimental dataindicate that independent residual analysis (IRA) method can provide efficient temporal features. Combining IRAfeatures with the CSP method, we obtain the optimal spatial and temporal features with which we achieve the bestclassification rate. The high classification rate indicates that the proposed method is promising for an EEG-basedbrain-computer interface.

Most brain-computer interfaces (BCIs) utilize univariate features (i.e., power spectrum or amplitude) for motor imagery (MI) pattern recognition, while less attention has been paid on multivariate analysis that considers information flow between various brain areas through brain connectivity estimations. Most recently, researches have proved that connectivity features were able to characterize different MI tasks. In this study, we investigated the performance of functional connectivity features measured by phase lag index (PLI), weighted phase lag index(wPLI) and phase-locking value(PLV) on MI classification. The widely-used filter-bank common spatial pattern (FBCSP) approach was employed here for performance assessment. A linear support vector machine was trained to classify different MI tasks using two publicly-available datasets from BCI Competition III and IV. The classification results showed that connectivity features achieved classification accuracy >85% in most cases and PLI outperformed all other methods including FBCSP. Our work suggested that functional connectivity features could be utilized as a powerful tool for recognizing different motor intention.

The spatial distribution of soil organic carbon (SOC) serves as critical geographic information for assessing ecosystem services and guides land management for migrating carbon emissions. Digital mapping of SOC is challenging due to the complex relationships between the soil and its environmental conditions. Except for the well-known topography and climate environmental covariates, the aboveground vegetation growth, which interacts with belowground soil carbon, influences SOC significantly over seasonal and interannual variations. Although several remote-sensing-based vegetation indices (e.g. NDVI and EVI) have been widely adopted in digital soil mapping, variables indicating long-term vegetation growth status have been less used. The vegetation phenology, an indicator of vegetation growth characteristics, can be used as a potential time series environmental covariate for SOC prediction. In this study, a CNN-RNN hybrid model was developed for SOC prediction with inputs of static and dynamic environmental variables in a study area located in Xuancheng City, China. The spatially contextual features in static variables (e.g., topographic variables) were extracted by the convolutional neural network (CNN), while the temporal features in dynamic variables (e.g., vegetation phenology over a long period) were extracted by a recurrent neural network (RNN) as represented by using a long short-term memory (LSTM) network. The ten-year phenological variables before the sampling year derived from satellite-based observations were adopted as new predictors reflecting historical temporal changes in vegetation in addition to the commonly used static variables. The random forest model was used as a reference model for comparison. Our results indicate that adding phenological variables can improve the soil carbon prediction accuracy, and demonstrate that the fine-tuned CNN-RNN model is potentially effective and can be a powerful model for SOC predictive mapping. We conclude that the hybrid deep learning models have great potential to enhance soil prediction by simultaneously extracting spatial and temporal latent features from different types of environmental variables, and highlight that using the long-term historical vegetation phenology information can serve as a useful extra input for future applications in the predictive mapping of soil carbon.ReferencesZhang, L., Cai, Y., Huang, H., Li, A., Yang, L., Zhou, C., 2022. A CNN-LSTM Model for Soil Organic Carbon Content Prediction with Long Time Series of MODIS-Based Phenological Variables. Remote Sensing 14, 4441Yang, L., Cai, Y., Zhang, L., Guo, M., Li, A., Zhou, C., 2021. A deep learning method to predict soil organic carbon content at a regional scale using satellite-based phenology variables. International Journal of Applied Earth Observation and Geoinformation 102, 102428He, X., Yang, L., Li, A., Zhang, L., Shen, F., Cai, Y., Zhou, C., 2021. Soil organic carbon prediction using phenological parameters and remote sensing variables generated from Sentinel-2 images. CATENA 205, 105442.

The canopy temperature of winter jujube serves as a direct indicator of plant water status and transpiration efficiency, making its accurate prediction a critical prerequisite for effective water management and optimized growth conditions in greenhouse environments. This study developed a data-driven model to forecast canopy temperature. The model serially integrates a Long Short-Term Memory (LSTM) network and a Random Forest (RF) algorithm, leveraging their complementary strengths in capturing temporal dependencies and robust nonlinear fitting. A three-stage framework comprising temporal feature extraction, multi-source feature fusion, and direct prediction was implemented to enable reliable nowcasting. Data acquisition and preprocessing were tailored to the greenhouse environment, involving multi-sensor data and thermal imagery processed with Robust Principal Component Analysis (RPCA) for dimensionality reduction. Key environmental variables were selected through Spearman correlation analysis. Experimental results demonstrated that the proposed LSTM–RF model achieved superior performance, with a determination coefficient (R2) of 0.974, mean absolute error (MAE) of 0.844 °C, and root mean square error (RMSE) of 1.155 °C, outperforming benchmark models including standalone LSTM, RF, Transformer, and TimesNet. SHAP (SHapley Additive exPlanations)-based interpretability analysis further quantified the influence of key factors, including the “thermodynamic state of air” driver group and latent temporal features, offering actionable insights for irrigation management. The model establishes a reliable, interpretable foundation for real-time water stress monitoring and precision irrigation control in protected winter jujube production systems.

Recently, Brain-Computer Interface (BCI) technology has been applied more and more in the field of clinical rehabilitation, which provides an effective way of communication for patients with brain disability and stroke. Because electroencephalogram (EEG) signals are extremely complex and contain many redundant signals, the recognition effect of BCI system based on EEG is not very well. The purpose of this paper is to solve the problem of brain intention classification in brain-computer interface system, which combines Deep Separation Convolutional Neural Network (DSCNN) and Gate Recurrent Unit (GRU) network to classify the motor imagination task of EEG. Firstly, one-dimensional timing EEG signals are transformed into two-dimensional array, and the temporal and spatial features of EEG signals are extracted by separate convolution. Then, these EEG signals containing spatio-temporal features are convolved once to extract spatial features, and the temporal features is extracted by GRU. Finally, the experiments shows that the final intention recognition accuracy reach 97.76% via the open physiological motor imagery data set EEGMMIDB, which is superior to some advanced research methods for motor imagery task recognition at present and helpful to restore the rehabilitation ability of patients with brain injury.

Network intrusion poses a severe threat to the Internet. Intrusion detection methods based on deep learning are very effective to process and analyze intrusion data. On the one hand, they use the Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) models to extract spatial and temporal features, respectively. However, they either adopt a single model or operate two models in series. And it fails to capture temporal-spatial features effectively. On the other hand, previous methods do not consider the multi-feature correlation of intrusion data. And then they cannot get better classification performance. To address the two above problems, we propose a hybrid neural network (HNN) model by integrating multi-feature correlation and temporal-spatial analysis. First, we adopt a contribution-based feature selection. Second, we reconstruct multi-feature correlation and then apply the CNN and LSTM in parallel to extract temporal-spatial features. Finally, we splice the temporal-spatial features with the correlation features, or study the influence of the temporal-spatial features by attention mechanism. Based on the above operations, we exploit the Deep Neural Network (DNN) to detect intrusion data. The experimental results show that HNN improves 3.78%, 1.31%, 0.21%, and 1.13% accuracy on the UNSW-NB15, AWID, CICIDS 2017, and CICIDS 2018 datasets.

Cognitive load refers to the mental effort required to process information and perform tasks, significantly influencing learning and performance outcomes. This paper presents a novel approach for cognitive load classification using a hybrid model that integrates Long Short-Term Memory (LSTM) networks with the Block Attention Module (BAM). Leveraging functional Near-Infrared Spectroscopy (fNIRS), we investigate the relationship between cognitive load and brain activity in a controlled experimental setting. Our methodology encompasses data collection from 50 participants engaged in various problem-solving tasks, with cognitive load categorized as high, medium, or low. The acquired fNIRS data underwent a rigorous preprocessing pipeline, including normalization and wavelet transform for feature extraction, enabling a comprehensive analysis of hemodynamic responses. The proposed model employs BAM to enhance feature representation by refining the importance of spatial and channel dimensions, thus improving the LSTM's ability to capture temporal dependencies in the data. The experimental results demonstrate significant performance improvements in cognitive load classification, showcasing the efficacy of the integrated LSTM-BAM architecture. This work not only contributes to the understanding of cognitive load dynamics but also highlights the potential of fNIRS as a non-invasive tool for real-time monitoring of cognitive performance, paving the way for advancements in instructional design and cognitive research.

In this work, efficient human activity recognition (HAR) algorithm based on deep learning architecture is proposed to classify activities into seven different classes. In order to learn spatial and temporal features from only 3D skeleton data captured from a “Microsoft Kinect” camera, the proposed algorithm combines both convolution neural network (CNN) and long short-term memory (LSTM) architectures. This combination allows taking advantage of LSTM in modeling temporal data and of CNN in modeling spatial data. The captured skeleton sequences are used to create a specific dataset of interactive activities; these data are then transformed according to a view invariant and a symmetry criterion. To demonstrate the effectiveness of the developed algorithm, it has been tested on several public datasets and it has achieved and sometimes has overcome state-of-the-art performance. In order to verify the uncertainty of the proposed algorithm, some tools are provided and discussed to ensure its efficiency for continuous human action recognition in real time.

Error-related potentials (ErrPs) have been proposed as a means for improving brain–computer interface (BCI) performance by either correcting an incorrect action performed by the BCI or label data for continuous adaptation of the BCI to improve the performance. The latter approach could be relevant within stroke rehabilitation where BCI calibration time could be minimized by using a generalized classifier that is continuously being individualized throughout the rehabilitation session. This may be achieved if data are correctly labelled. Therefore, the aims of this study were: (1) classify single-trial ErrPs produced by individuals with stroke, (2) investigate test–retest reliability, and (3) compare different classifier calibration schemes with different classification methods (artificial neural network, ANN, and linear discriminant analysis, LDA) with waveform features as input for meaningful physiological interpretability. Twenty-five individuals with stroke operated a sham BCI on two separate days where they attempted to perform a movement after which they received feedback (error/correct) while continuous EEG was recorded. The EEG was divided into epochs: ErrPs and NonErrPs. The epochs were classified with a multi-layer perceptron ANN based on temporal features or the entire epoch. Additionally, the features were classified with shrinkage LDA. The features were waveforms of the ErrPs and NonErrPs from the sensorimotor cortex to improve the explainability and interpretation of the output of the classifiers. Three calibration schemes were tested: within-day, between-day, and across-participant. Using within-day calibration, 90% of the data were correctly classified with the entire epoch as input to the ANN; it decreased to 86% and 69% when using temporal features as input to ANN and LDA, respectively. There was poor test–retest reliability between the two days, and the other calibration schemes led to accuracies in the range of 63–72% with LDA performing the best. There was no association between the individuals’ impairment level and classification accuracies. The results show that ErrPs can be classified in individuals with stroke, but that user- and session-specific calibration is needed for optimal ErrP decoding with this approach. The use of ErrP/NonErrP waveform features makes it possible to have a physiological meaningful interpretation of the output of the classifiers. The results may have implications for labelling data continuously in BCIs for stroke rehabilitation and thus potentially improve the BCI performance.

Temporal pattern-aware temperature forecasting using CatBoost: A hybrid machine learning approach

Advances in the field of Brain-Computer Interfaces (BCIs) aim, among other applications, to improve the movement capacities of people suffering from the loss of motor skills. The main challenge in this area is to achieve real-time and accurate bio-signal processing for pattern recognition, especially in Motor Imagery (MI). The significant interaction between brain signals and controllable machines requires instantaneous brain data decoding. In this study, an embedded BCI system based on fist MI signals is developed. It uses an Emotiv EPOC+ Brainwear®, an Altera SoCKit® development board, and a hexapod robot for testing locomotion imagery commands. The system is tested to detect the imagined movements of closing and opening the left and right hand to control the robot locomotion. Electroencephalogram (EEG) signals associated with the motion tasks are sensed on the human sensorimotor cortex. Next, the SoCKit processes the data to identify the commands allowing the controlled robot locomotion. The classification of MI-EEG signals from the F3, F4, FC5, and FC6 sensors is performed using a hybrid architecture of Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks. This method takes advantage of the deep learning recognition model to develop a real-time embedded BCI system, where signal processing must be seamless and precise. The proposed method is evaluated using k-fold cross-validation on both created and public Scientific-Data datasets. Our dataset is comprised of 2400 trials obtained from four test subjects, lasting three seconds of closing and opening fist movement imagination. The recognition tasks reach 84.69% and 79.2% accuracy using our data and a state-of-the-art dataset, respectively. Numerical results support that the motor imagery EEG signals can be successfully applied in BCI systems to control mobile robots and related applications such as intelligent vehicles.

Brain-computer interfaces (BCIs) are at the forefront of medical technology, with transformative potential for neurological rehabilitation [1]. Motor imagery-based braincomputer interfaces (MI-BCIs) hold significant clinical value in neurorehabilitation [2,3]. By facilitating direct communication between the brain and external devices, BCIs are broadening the scope of neurorehabilitation, particularly for individuals with conditions such as spinal cord injury (SCI), stroke, and unilateral spatial neglect (USN) [4][5][6]. This special topic aims to highlight recent advancements, diverse applications, and the complex challenges associated with BCIs in managing neurological disorders. It includes four notable contributions, each providing unique insights into the potential of BCIs to enhance quality of life and functional abilities for those affected.In the study by Guo et al. on spinal cord injury (SCI), the authors explored a novel approach combining epidural electrical stimulation (EES) with near-infrared nerve stimulation (nINS) to enhance motor function specificity in SCI rats. The results revealed that adding optical stimulation to conventional EES could selectively activate target muscles with reduced stimulation intensity, offering an effective strategy to mitigate unwanted muscle activation and improve movement control. This dualstimulation approach presents a promising new avenue for addressing motor dysfunctions associated with SCI, paving the way for more precise neuromodulation in rehabilitation settings [7].patients. Using graph theory, the study assessed cortical reorganization in individuals undergoing VR-assisted therapy. The results demonstrated that the VR-based rehabilitation tool, Gesture Collection, significantly enhanced functional connectivity within motor-related brain regions compared to standard therapy. The improvement in connectivity, particularly within the frontoparietal and somatosensory networks, highlights the potential of VR to promote neuroplasticity and facilitate motor recovery in stroke survivors. By providing engaging and targeted motor exercises, VR-assisted rehabilitation could serve as a valuable complement to traditional stroke recovery programs [8].Wang et al. investigated the neural characteristics associated with neuropathic pain and numbness in SCI patients. Using electroencephalography (EEG), the study identified distinct brain network patterns differentiating patients with neuropathic pain from those experiencing numbness. Notably, individuals suffering from pain exhibited reduced power in lower frequency bands (θ and α) and increased power in the higher frequency band (β), accompanied by altered network connectivity. These findings suggest that EEG-based metrics could serve as valuable biomarkers for distinguishing different neuropathic symptoms in SCI patients, thereby informing personalized intervention strategies to improve chronic pain management in this population [9].The article by Gouret et al. provides a focused review of brain-computer interface (BCI) applications in the context of unilateral spatial neglect (USN), a common yet often underrecognized consequence of stroke. The review emphasizes the limited use of BCIs in addressing cognitive deficits, particularly visuo-attentional impairments, and underscores the need for expanded research in this area. The authors propose integrating VR with BCIs as a rehabilitation tool for USN, leveraging the cognitive engagement provided by VR to support attentional recovery. This perspective advocates for broadening the scope of BCI applications beyond motor rehabilitation to include cognitive therapy, thereby addressing a significant gap in current neurological treatment approaches [10] Collectively, these studies highlight the versatility and potential of brain-computer interfaces (BCIs) in neurological rehabilitation. By enhancing motor recovery, managing pain, and supporting cognitive restoration, BCIs represent a multidimensional tool capable of addressing various aspects of neurological impairments [11]. However, several challenges remain, including the need for improved user accessibility, device usability, and ethical considerations. As BCI technology continues to advance, interdisciplinary collaboration will be crucial for translating experimental insights into real-world solutions that can be widely adopted in clinical practice. For instance, integrating optogenetic BCIs with functional imaging techniques could allow targeted modulation of specific circuits, further promoting neurorehabilitation [12][13].This special issue aims to inspire ongoing research and innovation in the field of BCIs, with the ultimate goal of empowering individuals with neurological disorders to lead more independent and fulfilling lives. Through continued exploration and application, BCIs hold the promise of bridging the gap between neurological impairment and functional recovery, offering hope to patients, clinicians, and researchers alike.