Search Results (1,345)

In the semi-desert aeolian sand areas of Northern China, surface deformation monitoring with SAR is challenged by loss of coherence due to mobile dunes, seasonal vegetation changes, and large-gradient, nonlinear subsidence from underground mining. This study utilizes PALSAR-2 (L-band, 3 m resolution) and Sentinel-1 (C-band, 30 m resolution) data, applying InSAR and Offset tracking methods combined with differential, Stacking, and SBAS techniques to analyze deformation monitoring effectiveness and propose an efficient dynamic monitoring strategy for the Shendong Coalfield. The main conclusions can be summarized as follows: (1) PALSAR-2 data, which has advantages in wavelength and resolution (L-band, multi-look spatial resolution of 3 m), exhibits better interference effects and deformation details compared to Sentinel-1 data (C-band, multi-look spatial resolution of 30 m). The highly sensitive differential-InSAR (D-InSAR) can promptly detect new deformations, while Stacking-InSAR can accurately delineate the range of rock strata movement. SBAS-InSAR can reflect the dynamic growth process of the deformation range as a whole, and SBAS-Offset is suitable for observing the absolute values and morphology of the surface moving basin. The combined application of Stacking-InSAR and Stacking-Offset methods can accurately acquire the three-dimensional deformation field of mining-induced strata movement. (2) The spatiotemporal process of surface deformation caused by coal mining-induced strata movement revealed by InSAR exhibits good correspondence with both the underground mining progress and the development of ground fissures identified in UAV images. (3) The maximum displacement along the line of sight (LOS) measured in the mining area is approximately 2 to 3 m, which is close to the 2.14 m observed on site and aligns with previous studies. The calculated advance influence angle of the No. 22308 working face in the study area is about 38.3°. The influence angle on the solid coal side is 49°, while that on the goaf side approaches 90°. These findings further deepen the understanding of rock movement and surface displacement parameters in this region. The dynamic monitoring strategy proposed in this study is cost-effective and operational, enhancing the observational effectiveness of InSAR technology for surface deformation due to coal mining in this area, and it enriches the understanding of surface strata movement patterns and parameters in this region. Full article

Steady-state visual evoked potential (SSVEP)-based brain–computer interface (BCI) technology offers a promising solution for wheelchair control by translating neural signals into navigation commands. A major challenge—signal noise caused by eye blinks—is addressed in this feasibility study through real-time blink detection and correction. The proposed design utilizes sensors to capture both SSVEP and blink signals, enabling the isolation and compensation of interference, which improves control accuracy by 14.68%. Real-time correction during blinks significantly enhances system reliability and responsiveness. Furthermore, user data and global positioning system (GPS) trajectories are uploaded to the cloud via Wi-Fi 6E for continuous safety monitoring. This approach not only restores mobility for users with physical disabilities but also promotes independence and spatial autonomy. Full article

Accurate traffic prediction is essential for intelligent transportation systems, urban mobility management, and traffic optimization. However, existing deep learning approaches often struggle to jointly capture complex spatial dependencies and temporal dynamics, and they are prone to overfitting when modeling large-scale traffic networks. To address these challenges, we propose the GSF-LLM (graph-enhanced spatio-temporal fusion-based large language model), a novel framework that integrates large language models (LLMs) with graph-based spatio-temporal learning. GSF-LLM employs a spatio-temporal fusion module to jointly encode spatial and temporal correlations, combined with a partially frozen graph attention (PFGA) mechanism to model topological dependencies while mitigating overfitting. Furthermore, a low-rank adaptation (LoRA) strategy is adopted to fine-tune a subset of LLM parameters, improving training efficiency and generalization. Experiments on multiple real-world traffic datasets demonstrate that GSF-LLM consistently outperforms state-of-the-art baselines, showing strong potential for extension to related tasks such as data imputation, trajectory generation, and anomaly detection. Full article

The application of orchard inspection robots has become increasingly widespread. How-ever, achieving autonomous navigation in unstructured environments continues to pre-sent significant challenges. This study investigates the Simultaneous Localization and Mapping (SLAM) navigation system of an orchard inspection robot and evaluates its performance using Light Detection and Ranging (LiDAR) technology. A mobile robot that integrates tightly coupled multi-sensors is developed and implemented. The integration of LiDAR and Inertial Measurement Units (IMUs) enables the perception of environmental information. Moreover, the robot’s kinematic model is established, and coordinate transformations are performed based on the Unified Robotics Description Format (URDF). The URDF facilitates the visualization of robot features within the Robot Operating System (ROS). ROS navigation nodes are configured for path planning, where an improved A* algorithm, combined with the Dynamic Window Approach (DWA), is introduced to achieve efficient global and local path planning. The comparison of the simulation results with classical algorithms demonstrated the implemented algorithm exhibits superior search efficiency and smoothness. The robot’s navigation performance is rigorously tested, focusing on navigation accuracy and obstacle avoidance capability. Results demonstrated that, during temporary stops at waypoints, the robot exhibits an average lateral deviation of 0.163 m and a longitudinal deviation of 0.282 m from the target point. The average braking time and startup time of the robot at the four waypoints are 0.46 s and 0.64 s, respectively. In obstacle avoidance tests, optimal performance is observed with an expansion radius of 0.4 m across various obstacle sizes. The proposed combined method achieves efficient and stable global and local path planning, serving as a reference for future applications of mobile inspection robots in autonomous navigation. Full article

Memory forensics is an essential cybersecurity tool that comprehensively examines volatile memory to detect the malicious activity of fileless malware that can bypass disk analysis. Image-based detection techniques provide a promising solution by visualizing memory data into images to be used and analyzed by image processing tools and machine learning methods. However, the effectiveness of image-based data for detection and classification requires high computational efforts. This paper investigates the efficacy of texture-based methods in detecting and classifying memory-resident or fileless malware using different image resolutions, identifying the best feature descriptors, classifiers, and resolutions that accurately classify malware into specific families and differentiate them from benign software. Moreover, this paper uses both local and global descriptors, where local descriptors include Oriented FAST and Rotated BRIEF (ORB), Scale-Invariant Feature Transform (SIFT), and Histogram of Oriented Gradients (HOG) and global descriptors include Discrete Wavelet Transform (DWT), GIST, and Gray Level Co-occurrence Matrix (GLCM). The results indicate that as image resolution increases, most feature descriptors yield more discriminative features but require higher computational efforts in terms of time and processing resources. To address this challenge, this paper proposes a novel approach that integrates Local Interpretable Model-agnostic Explanations (LIME) with deep learning models to automatically identify and crop the most important regions of memory images. The LIME’s ROI was extracted based on ResNet50 and MobileNet models’ predictions separately, the images were resized to 128 × 128, and the sampling process was performed dynamically to speed up LIME computation. The ROIs of the images are cropped to new images with sizes of (100 × 100) in two stages: the coarse stage and the fine stage. The two generated LIME-based cropped images using ResNet50 and MobileNet are fed to the lightweight neural network to evaluate the effectiveness of the LIME-based identified regions. The results demonstrate that the LIME-based MobileNet model’s prediction improves the efficiency of the model by preserving important features with a classification accuracy of 85% on multi-class classification. Full article

Lameness remains a significant welfare and productivity challenge in dairy farming, often underdiagnosed due to the limitations of conventional detection methods. Unlike most previous approaches to lameness detection that rely on a single-sensor or gait-based measurement, this study integrates four complementary data domains—behavioural, physiological, biochemical, and milk composition parameters—collected from 272 dairy cows during early lactation to enhance diagnostic accuracy and biological interpretability. The main objective of this study was to evaluate and compare the diagnostic classification performance of multiple machine learning (ML) algorithms trained on multimodal data collected at the time of clinical lameness diagnosis during early lactation, and to identify the most influential physiological and biochemical traits contributing to classification accuracy. Specifically, six algorithms—random forest (RF), neural network (NN), Ensemble, support vector machine (SVM), k-nearest neighbors (KNN), and logistic regression (LR)—were assessed. The input dataset integrated physiological parameters (e.g., water intake, body temperature), behavioural indicators (rumination time, activity), blood biochemical biomarkers (non-esterified fatty acids (NEFA), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), gamma-glutamyl transferase (GGT)), and milk quality traits (fat, protein, lactose, temperature). Among all models, RF achieved the highest validation accuracy (97.04%), perfect validation specificity (100%), and the highest normalized Matthews correlation coefficient (nMCC = 0.94), as determined through Monte Carlo cross-validation on independent validation sets. Lame cows showed significantly elevated NEFA and body temperatures, reflecting enhanced lipid mobilization and inflammatory stress, alongside reduced water intake, milk protein, and lactose content, indicative of systemic energy imbalance and impaired mammary function. These physiological and biochemical deviations emphasize the multifactorial nature of lameness. Linear models like LR underperformed, likely due to their inability to capture the non-linear and interactive relationships among physiological, biochemical, and milk composition features, which were better represented by tree-based and neural models. Overall, the study demonstrates that combining sensor data with blood biomarkers and milk traits using advanced ML models provides a powerful, objective tool for the clinical classification of lameness, offering practical applications for precision livestock management by supporting early, data-driven decision-making to improve welfare and productivity on dairy farms. Full article

Machine harvesting of cotton in Xinjiang has significantly improved harvesting efficiency; however, it has also resulted in a considerable increase in residual mulch content within the cotton, which has severely affected the quality and market value of cotton textiles. Existing mulch detection algorithms based on machine vision generally suffer from complex parameterization and insufficient real-time performance. To overcome these limitations, this study proposes a novel mulch detection algorithm, Mulch-YOLO, developed on the YOLOv11 framework. Specifically, an improved CBAM (Convolutional Block Attention Module) is incorporated into the BiFPN (Bidirectional Feature Pyramid Network) to achieve more effective fusion of multi-scale mulch features. To enhance the semantic representation of mulch features, a modified Content-Aware ReAssembly of Features module, CARAFE-Mulch (Content-Aware ReAssembly of Features), is designed to reorganize feature maps, resulting in stronger feature expressiveness compared with the original representations. Furthermore, the MobileOne module is optimized by integrating the DECA Dilated Efficient Channel Attention (Dilated Efficient Channel Attention) module, thereby reducing both the parameter count and computational load while improving detection efficiency in real time. To verify the effectiveness of the proposed approach, experiments were conducted on a real-world dataset containing 20,134 images of low-visual-saliency plastic mulch. The results indicate that Mulch-YOLO achieves a lightweight architecture and high detection accuracy. Compared with YOLOv11n, the proposed method improves mAP@0.5 by 4.7% and mAP@0.5:0.95 by 3.3%, with a 24% reduction in model parameters. Full article

Background: Over the past several decades, the prevalence of gestational diabetes mellitus (GDM) has risen markedly worldwide, posing serious threats to both maternal and child health by increasing adverse pregnancy outcomes and long-term metabolic risks. Developing effective risk prediction tools for early detection and intervention has become the most important clinical priority in this field. The current GDM prediction models primarily rely on non-modifiable factors, for example age and body mass index, while modifiable factors such as lifestyle and health literacy, although strongly associated with GDM, have not been fully utilized in risk assessment. This study sought to establish and validate a nomogram prediction model combining modifiable and non-modifiable risk factors, with the goal of identifying high-risk Chinese pregnant women with GDM at an early stage and promoting targeted prevention and personalized prenatal management. Methods: A multicenter study was conducted across 7 maternal health institutions in Southern China (2021–2023), enrolling 806 singleton pregnant women (14–23⁺⁶ weeks). The collected data included sociodemographic, clinical history, and modifiable factors collected through validated questionnaires: dietary quality, physical activity level, sleep quality, and nutrition and health literacy. GDM was diagnosed via 75 g oral glucose tolerance test at 24–28 weeks. Predictive factors were identified through multi-variable logistic regression. A nomogram model was developed (70% modeling group) and validated (30% validation group). Receiver operator characteristic curves, calibration curves, and decision curve analysis were used to evaluate the prediction ability, the degree of calibration, and the clinical benefit of the model, respectively. Results: The finalized risk prediction model included non-modifiable factors such as maternal age, pre-pregnancy weight, and maternal polycystic ovary syndrome, as well as modifiable factors including dietary quality, physical activity level, sleep quality, nutrition and health literacy. The application of the nomogram in the modeling group and the validation groups showed that the model had high stability, favorable predictive ability, good calibration effect and clinical practicality. Conclusions: Overall, the integrated model demonstrates significant clinical utility as it facilitates the prompt identification of individuals at heightened risk and offers actionable targets for personalized interventions. In terms of future implementation, this model can be integrated into prenatal care as a rapid scoring table during early pregnancy consultations or incorporated into mobile health applications. This approach fosters precise prevention strategies for GDM in maternal health by emphasizing nutrition and health literacy, supplemented by coordinated adjustments in diet, physical activity, and sleep. Full article

Enhancing road safety requires an accurate assessment of the drivers’ mental states. The Driver Status Test App (DSTA) is designed to detect conditions such as intoxication, fatigue, and cognitive impairment in simulated driving environments. Utilizing a dual-platform approach, it integrates mobile data collection via React Native and Firebase with web-based management using React and TypeScript. The mobile application conducts real-time assessments of cognitive and motor functions, while the web interface offers data visualization, trend analysis, and results exportation. DSTA evaluates driver impairment through metrics such as tracking, precision, balance, and choice reaction, producing an objective impairment score. These assessments are rapid, scalable, and adaptable for various research and regulatory purposes. The composite scoring framework differentiates between impaired and unimpaired states, making DSTA valuable for driver training programs, regulatory assessments, and autonomous vehicle research, where monitoring human factors is crucial. Full article

Attention and brain focus are essential in human activities that require learning. In higher education, a popular means of acquiring knowledge and information is through serious games. The need for integrating digital learning tools, including serious games, into university curricula has been demonstrated by the students’ preferences that are oriented more towards engaging and interactive alternatives than traditional education. This study examines real-time attention measurement in serious games using wearable brain–computer interfaces (BCIs). By capturing electroencephalography (EEG) signals non-invasively, the system continuously monitors players’ cognitive states to assess attention levels during gameplay. The novel approach proposes adaptive attention measurements to investigate the ability to maintain attention during cognitive tasks of different durations and intensities, using a single-channel EEG system—NeuroSky Mindwave Mobile 2. The measures have been achieved on ten volunteer master’s students in Computer Science. Attention levels during short and intense tasks were compared with those recorded during moderate and long-term activities like watching an educational lecture. The aim was to highlight differences in mental concentration and consistency depending on the type of cognitive task. The experiment was designed following a unique protocol applied to all ten students. Data were acquired using the NeuroExperimenter software 6.6, and analytics were performed in RStudio Desktop for Windows 11. Data is available at request for further investigations and analytics. Experimental results demonstrate that wearable BCIs can reliably detect attention fluctuations and that integrating this neuroadaptive feedback significantly enhances player focus and immersion. Thus, integrating real-time cognitive monitoring in serious game design is an efficient method to optimize cognitive load and create personalized, engaging, and effective learning or training experiences. Beta and attention brain waves, associated with concentration and mental processing, had higher values during the gameplay phase than in the lecture phase. At the same time, there are significant differences between participants—some react better to reading, while others react better to interactive games. The outcomes of this study contribute to the design of personalized learning experiences by customizing learning paths. Integrating NeuroSky or similar EEG tools can be a significant step toward more data-driven, learner-aware environments when designing or evaluating educational games. Full article

In the last decade, the incorporation of Artificial Intelligence (AI) with autonomous vehicles (AVs) has transformed transportation, mobility, and smart mobility systems. The present study provides a systematic review of global trends, applications, and challenges at the intersection of AI, including Machine Learning (ML), Deep Learning (DL), and autonomous vehicle technologies. Using data extracted from Clarivate Analytics’ Web of Science Core Collection and a set of specific keywords related to both AI and autonomous (electric) vehicles, this paper identifies the themes presented in the scientific literature using thematic maps and thematic map evolution analysis. Furthermore, the research topics are identified using both thematic maps, as well as Latent Dirichlet Allocation (LDA) and BERTopic, offering a more faceted insight into the research field as LDA enables the probabilistic discovery of high-level research themes, while BERTopic, based on transformer-based language models, captures deeper semantic patterns and emerging topics over time. This approach offers richer insights into the systematic review analysis, while comparison in the results obtained through the various methods considered leads to a better overview of the themes associated with the field of AI in autonomous vehicles. As a result, a strong correspondence can be observed between core topics, such as object detection, driving models, control, safety, cybersecurity and system vulnerabilities. The findings offer a roadmap for researchers and industry practitioners, by outlining critical gaps and discussing the opportunities for future exploration. Full article

The rapid evolution of 5G and emerging 6G networks has increased system complexity, data volume, and security risks, making anomaly detection vital for ensuring reliability and resilience. However, existing machine learning (ML)-based approaches still face challenges related to poor generalization, weak temporal modeling, and degraded accuracy under heterogeneous and imbalanced real-world conditions. To overcome these limitations, a hybrid time series transformer–deep belief network (HTST-DBN) is introduced, integrating the sequential modeling strength of TST with the hierarchical feature representation of DBN, while an improved orchard algorithm (IOA) performs adaptive hyper-parameter optimization. The framework also embodies the concept of symmetry and asymmetry. The IOA introduces controlled symmetry-breaking between exploration and exploitation, while the TST captures symmetric temporal patterns in network traffic whose asymmetric deviations often indicate anomalies. The proposed method is evaluated across four benchmark datasets (ToN-IoT, 5G-NIDD, CICDDoS2019, and Edge-IoTset) that capture diverse network environments, including 5G core traffic, IoT telemetry, mobile edge computing, and DDoS attacks. Experimental evaluation is conducted by benchmarking HTST-DBN against several state-of-the-art models, including TST, bidirectional encoder representations from transformers (BERT), DBN, deep reinforcement learning (DRL), convolutional neural network (CNN), and random forest (RF) classifiers. The proposed HTST-DBN achieves outstanding performance, with the highest accuracy reaching 99.61%, alongside strong recall and area under the curve (AUC) scores. The HTST-DBN framework presents a scalable and reliable solution for anomaly detection in next-generation mobile networks. Its hybrid architecture, reinforced by hyper-parameter optimization, enables effective learning in complex, dynamic, and heterogeneous environments, making it suitable for real-world deployment in future 5G/6G infrastructures. Full article

Echocardiography remains a vital part of the initial assessment and monitoring of oncological patients. It allows for proper treatment selection but can also reveal life-threatening complications, including impaired left ventricular function or thromboembolism. It can rarely detect intracardiac masses that require further investigation. In the presented case, a 51-year-old female patient with left-sided breast cancer, who had undergone neoadjuvant chemotherapy, was hospitalised due to a right atrial mass identified via routine transthoracic echocardiography (TTE). Initial anticoagulation therapy showed no clinical improvement. Follow-up TTE revealed a 12 × 19 mm hyperechogenic, mobile mass in the right atrium (RA). Computed tomography angiography (CTA) ruled out pulmonary embolism and revealed that the mass was located close to the tip of the vascular access port. Transoesophageal echocardiography showed that the lesion was not connected to the vascular port. Based on location and mobility, the lesion was most consistent with a cardiac myxoma. After the Heart Team made a decision, endovascular intervention using a vacuum-assisted device was performed without complications. Histopathological examination excluded thrombosis and myxoma, revealing a fibro-inflammatory lesion. A multimodality approach is necessary to assess RA masses. However, even an extensive evaluation could be misleading, so treatment options should always be subject to the Heart Team’s decision. Full article

The transition to 5G networks brings unprecedented speed, ultra-low latency, and massive connectivity. Nevertheless, it introduces complex traffic patterns and broader attack surfaces that render traditional intrusion detection systems (IDSs) ineffective. Existing rule-based methods and classical machine learning approaches struggle to capture the temporal and dynamic characteristics of 5G traffic, while many deep learning models lack interpretability, making them unsuitable for high-stakes security environments. To address these challenges, we propose Bidirectional Temporal Anomaly Detector (BiTAD), a deep temporal learning architecture for anomaly detection in 5G networks. BiTAD leverages dual-direction temporal sequence modelling with attention to encode both past and future dependencies while focusing on critical segments within network sequences. Like many deep models, BiTAD’s faces interpretability challenges. To resolve its “black-box” nature, a dual-perspective explainability module, coined TwinLens, is proposed. This module integrates SHAP and TimeSHAP to provide global feature attribution and temporal relevance, delivering dual-perspective interpretability. Evaluated on the public 5G-NIDD dataset, BiTAD demonstrates superior detection performance compared to existing models. TwinLens enables transparent insights by identifying which features and when they were most influential to anomaly predictions. By jointly addressing the limitations in temporal modelling and interpretability, our work contributes a practical IDS framework tailored to the demands of next-generation mobile networks. Full article

The growing abundance of traffic data offers new opportunities to uncover dynamic traffic patterns in urban road networks, providing valuable insights for promoting sustainable mobility. By leveraging these data, road segments can be grouped into communities to capture the spatiotemporal correlations driving the dynamic evolution of traffic states. However, existing distance-based methods lack the capacity to facilitate multi-scale analysis of urban traffic patterns and are limited in capturing the heterogeneity of road regions. To address this gap, in this study, we introduce a traffic-data-driven approach to detect road segment communities and extract multi-scale traffic patterns. Here, traffic data are mapped onto a dual graph of urban road networks, with node correlations weighted using Dynamic Time Warping (DTW). A hierarchical community detection algorithm is then applied to identify multi-scale communities, revealing the spatiotemporal structure of urban traffic dynamics. The robustness and effectiveness of the proposed method were tested on the road network of Chengdu. The results show that the method successfully integrates the topological structure with traffic data, capturing multi-scale spatial autocorrelation communities. By characterizing the evolution of traffic patterns, our method has potential applications in traffic prediction, traffic control, and urban planning applications, contributing to sustainable urban transportation through congestion mitigation and efficiency enhancement. Full article