Search Results (492)

Coal spontaneous combustion in gob areas is a major disaster endangering safe production in underground coal mines, and accurate prediction of carbon monoxide (CO), the core signature gas of coal oxidation, is critical for early warning and targeted prevention of mine fire disasters. However, CO concentration in gob areas is governed by complex gas–solid thermal–chemical multi-field coupling, presenting strong nonlinear characteristics. Traditional numerical methods suffer from prohibitive computational cost, purely data-driven models have inherent black-box defects, and conventional Physics-Informed Neural Networks (PINNs) require explicit full governing equations, which are hard to establish for such complex systems. This paper first proposes a Physics-Informed Modified Kolmogorov–Arnold Network (PIM-KAN), which deeply integrates domain physical knowledge with KAN architecture via a physics encoding layer, a residual-modified KAN layer, a multi-physics attention mechanism, and a multi-term physical consistency constraint framework. Experiments on 3125 real coal mine field samples show that the PIM-KAN achieves R² = 0.9965 and RMSE = 0.9290 ppm, reducing RMSE by 19.5% compared with MLP, and outperforming all baseline models. Ablation studies confirm the significant contribution of each innovation module, and attention weight analysis is highly consistent with Arrhenius reaction kinetics, verifying its superior prediction accuracy, physical consistency and intrinsic interpretability. Full article

Reliable fault detection and quantification are essential for the operational integrity of electric machines. While traditional current-based analysis relies on harmonic signatures or wavelet-based time-frequency representations, this study investigates modern learning formulations that capture spectral, multiscale, and temporal characteristics of fault-affected signals. Moving beyond conventional models, including our earlier CNN-based approaches, we develop sequence-based and operator-learning architectures within a multi-output formulation for eccentricity fault analysis. Three models are investigated: Mamba for temporal dynamics, the Fourier Neural Operator for global spectral mapping, and the Wavelet Neural Operator for localized multiscale decomposition. Evaluated on induction, salient pole synchronous, and inverter-based reluctance synchronous machines, each model maps stator current waveforms to multiple diagnostic quantities, including voltages, operating conditions, and fault severity. With time-delay embedding, all three achieve low prediction errors, with severity RMSE reaching the ${10}^{-4}$ scale for the induction machine, a notable reduction from the $0.04$ errors of our earlier hierarchical CNN models. These results show that modern sequence-based and operator-learning formulations can broaden machine fault analysis by enabling simultaneous prediction and estimation of multiple aspects of machine condition within a single model. Full article

As modern networks expand, the volume and destructiveness of cyberattacks continue to escalate, necessitating effective defense mechanisms. Intrusion Detection Systems (IDSs) are critical for maintaining network security; however, traditional signature-based systems often fail to detect zero-day attacks. This study explores recent advancements in Deep Learning (DL) for cybersecurity by analyzing and replicating the “IDS-ML” framework, an open-source repository for IDS development. We evaluate the performance of five deep learning Convolutional Neural Network (CNN) architectures adapted for intrusion detection via transfer learning on the CICIDS2017 dataset, and propose an enhancement by integrating Automated Machine Learning (AutoML) techniques that achieves a 94.7% reduction in model parameters while maintaining comparable accuracy, thus making our enhanced models suitable for deployment on edge devices. We further validate deployment feasibility by benchmarking both the baseline InceptionV3 and AutoML models on a Raspberry Pi 4, demonstrating an 18.7× inference speedup and 3.5× CPU reduction, with no change in predicted classes from model conversion. Our results confirm that lightweight AutoML architectures enable practical “zero-touch” edge-based intrusion detection on resource-constrained hardware. Full article

Background/Objective: Forensic document examination (FDE) traditionally relies on subjective expert opinion. This preliminary study was designed to develop and validate a hybrid deep learning model (ResNet-50 + bidirectional long short-term memory [BiLSTM]) for quantitative handwriting and signature feature analysis, and to compare its performance, under standardized experimental conditions, with that of three certified forensic document examiners. Methods: Handwriting and signature samples were collected from 225 individuals in a standardized setting. Fifteen quantitative handwriting features were extracted, the dataset was split into training (70%, n = 158) and testing (30%, n = 67) subsets using stratified random sampling, and ground truth for analytic categories was defined by majority consensus among the three examiners (with adjudicated review for disagreements). A hybrid architecture combining a ResNet-50 backbone and a bidirectional LSTM encoder was used. Results: The model demonstrated 93.4% accuracy, an F1-score of 0.926, and an AUC-ROC of 0.968 on the held-out test set. Under our task-specific experimental conditions, the model performed better than examiners on slant analysis (96.8% vs. 93.2%, p = 0.002), pressure profiling (94.1% vs. 91.7%, p = 0.019), and age estimation (87.4% vs. 82.1%, p = 0.011); examiners performed better on forgery detection (95.8% vs. 91.2%, p = 0.008) and signature verification (96.1% vs. 92.3%, p < 0.012). Mean processing time was reduced by 99.6% (0.8 s vs. 197 s per case). Conclusions: Within the limits of this preliminary single-centre study, the system showed performance comparable to certified examiners on several quantitative tasks and complementary strengths overall, supporting its feasibility as an adjunctive tool in a hybrid human–AI workflow. Broader, multi-centre validation and explainability work are required before any forensic deployment can be considered. Full article

Barely Visible Impact Damage (BVID) in heterogeneous polymer-matrix composites remains difficult to detect because subtle damage signatures are often masked by complex architectures, hybrid textures, and overlapping failure morphologies. This study therefore presents an experimentally grounded, physics-aware, and statistically validated vision-based inspection framework rather than a purely detector-centered benchmarking exercise. Real post-impact images were obtained from controlled low-velocity impact experiments on 20 composite architectures and 60 physical specimens, yielding approximately 2000 images across laminated, hybrid, textile-reinforced, and sandwich structures. The dataset was organized using a specimen-disjoint splitting protocol to prevent leakage across training, validation, and test subsets. To improve robustness while preserving physical realism, a physically grounded Albumentations strategy was developed using only physically admissible transformations and explicit exclusion of non-physical operations that could distort damage morphology or surface continuity. Model development was further complemented by a hybrid hardware workflow in which cloud-based GPU training was combined with deployment-oriented inference profiling on resource-constrained edge-like hardware, thereby linking detection accuracy to practical industrial feasibility. In addition, model performance was evaluated under a standardized training budget and validated through repeated runs, Friedman significance testing, and Holm-corrected Wilcoxon signed-rank pairwise comparisons to ensure error-controlled interpretation of inter-model differences. Across the evaluated compact YOLO families, YOLO26s delivered the strongest overall performance, reaching 0.841 mAP@0.5, 0.586 ± 0.004 mAP@0.5:0.95, and an F1-score of 0.809, while YOLO11s achieved the highest precision and YOLO26n remained competitive in recall with nano-level compactness. Overall, the results show that experimentally generated heterogeneous composite data, morphology-preserving augmentation strategy development, leakage-aware dataset design, deployment-oriented computational profiling, and statistically grounded validation together provide a more robust and application-relevant basis for automated BVID detection in polymer-matrix composite structures. Full article

Systemic lupus erythematosus (SLE) is a prototypic systemic autoimmune disorder arising from the convergence of genetic susceptibility, epigenetic remodeling, environmental exposures, and dysregulated immune networks. Although traditionally characterized by autoantibody production and immune complex mediated tissue injury, advances in genomics, systems immunology, and multi-omics profiling have revealed that lupus represents a multilayered failure of immune homeostasis driven by interconnected molecular circuits. Genetic variants enriched in regulatory immune enhancers establish a permissive transcriptional landscape that sensitizes innate nucleic acid sensing pathways and interferon signaling. Epigenetic remodeling further amplifies inflammatory transcriptional programs, while environmental triggers such as ultraviolet radiation and viral infection initiate bursts of nucleic acid release and immune activation. Defective apoptotic cell clearance, mediated in part by scavenger receptor dysfunction and complement abnormalities, increases the availability of immunogenic nucleic acids that engage pattern recognition receptors and drive chronic type I interferon production. This interferon-dominated environment rewires immune cell metabolism, alters differentiation trajectories of T and B lymphocytes, and sustains autoreactive immune circuits. Emerging multi-omics studies reveal distinct molecular endotypes defined by interferon signatures, metabolic states, and immune cell composition, highlighting the heterogeneity of disease mechanisms across patients. In this review, we integrate genetic, epigenetic, metabolic, and immunological insights to propose a systems-level model of lupus pathogenesis in which defective debris clearance, nucleic acid sensing, interferon amplification, and metabolic reprograming form a self-reinforcing pathogenic network. Understanding this integrated molecular architecture provides a foundation for biomarker-guided therapeutic strategies and precision medicine approaches aimed at disrupting the key nodes that sustain chronic autoimmunity in SLE. Full article

Background/Objectives: Escherichia coli (E. coli) spans commensal, intestinal pathogenic, and extraintestinal pathogenic lineages distributed across human, animal, and environmental reservoirs, yet the extent to which virulence architectures are shared across these compartments remains incompletely understood. Using a One Health framework, we profiled putative virulence determinants in pooled multidrug-resistant (MDR) E. coli source groups representing human, animal, and environmental sectors. Methods: Virulence genes were predicted with VirulenceFinder, and presence–absence profiles were integrated to define functional composition, sector overlap, source-group distribution breadth, and pathotype-associated signatures. Predicted pathogenic potential was assessed with PathogenFinder and compared with pathogenic family richness. Results: Overall, 114 putative virulence genes were detected, with adhesion/colonization functions dominating the virulome (33/114), followed by toxin-associated genes (12/114). A conserved core of 50 virulence genes was shared across all three sectors, including determinants linked to serum resistance (iss, ompT, traT), adhesion (csgA, fimH), stress adaptation (terC), and iron acquisition (sitA, iutA, fyuA). ExPEC-associated determinants were most numerous in environmental source groups (n = 52), whereas diarrheagenic E. coli markers were most frequent in animal-associated groups (n = 42). LEE-associated effectors were infrequent and largely absent from human source groups. Despite ecological differences in virulence composition, pathogenicity scores remained consistently high across sectors (0.83–0.92) and showed no significant association with pathogenic family richness (Spearman’s ρ = 0.197, p = 0.392). Conclusions: Within the limits of pooled source-group analysis, these findings suggest that MDR E. coli across One Health compartments shares a broadly distributed, ExPEC-associated virulence repertoire overlaid with sector-specific pathotype signals, underscoring the value of integrated genomic surveillance while highlighting the need for isolate-resolved analysis. Full article

Reliable perception of small floating objects is critical for the management of aquatic environments and supports key applications, including the autonomous navigation of Unmanned Surface Vehicles (USVs), waterborne debris monitoring, and Search and Rescue (SAR) operations. However, accurate detection in dynamic water-surface environments remains a significant challenge, as targets are frequently obscured by high-frequency wave clutter, and feature distributions are destabilized by covariate shifts caused by illumination. To address these limitations, this study proposes YOLO11-MCN, a real-time detection framework that integrates two architectural components specifically designed for water-surface monitoring. The Multi-Scale Contextual Attention (MSCA) module distinguishes target signatures from background noise by aggregating contextual information across heterogeneous receptive fields, thereby suppressing false positives generated by waves. The Channel Normalization Attention Mechanism (CNAM) addresses illumination instability through feature statistic calibration based on Group Normalization, effectively mitigating covariate shifts induced by extreme lighting variations. Furthermore, these components are complemented by a high-resolution P2 detection head, which recovers the geometric details of small-scale targets typically lost during downsampling. Extensive experiments conducted on a dataset of 5812 images demonstrate that YOLO11-MCN achieves an mAP@0.5 of 92.7%, outperforming the YOLO11n baseline by 5.9 percentage points. Robustness evaluations confirm that MSCA and CNAM significantly reduce missed detections under severe wave clutter and backlighting conditions. With a recall of 90.5%, an inference speed of 94 FPS on desktop hardware, and a compact footprint of 3.89M parameters and 14.8 GFLOPs, the proposed framework offers a robust and efficient solution for intelligent water-surface surveillance systems within the single-class detection paradigm evaluated in this study, with strong potential for edge-device deployment following platform-specific optimization. Full article

Historically, critical water infrastructures have operated with limited digitalization, relying on legacy protocols designed without intrinsic security. The rapid integration of advanced IoT telemetry into Operational Technology (OT) networks has dissolved traditional air gaps, exposing these facilities to severe cyber–physical threats. Concurrently, regulatory frameworks such as the European NIS2 Directive and the Cyber Resilience Act (CRA) now strictly mandate robust risk monitoring for essential entities. To address these challenges, this study develops a non-intrusive, hybrid Intrusion Detection System (IDS) tailored for converged IT/OT environments. Engineered upon the Snort 3 multi-threaded engine, the architecture captures both North–South and East–West traffic. A defense-in-depth rule set was constructed using threat intelligence (MITRE ATT&CK, CISA KEV) to perform Deep Packet Inspection (DPI) across legacy industrial protocols (Modbus, S7Comm, CIP) and IoT application layers (MQTT, HTTP). Experimental validation against high-volume synthetic packet captures (exceeding 170,000 packets) replicating specific manufacturer vulnerabilities (CVEs) demonstrated an improvement in the detection rate from a 0% baseline to 100%. Crucially, the system demonstrated high scalability and minimal computational overhead, processing high-volume traffic streams with zero dropped packets. This contextualized signature approach provides the deterministic security required to ensure operational continuity and regulatory compliance in modern water infrastructures. Full article

The rapid proliferation of Internet of Things (IoT) devices in cloud-based e-learning platforms has posed significant security risks, particularly in protecting learner information, authentication of devices, and safe communication in the highly heterogeneous learning settings. Current cryptographic solutions are largely based on classical public-key infrastructure (PKI) protocols such as RSA and ECC, which will become vulnerable with the advent of large-scale quantum computers capable of executing Shor’s algorithm. In addition, traditional perimeter-based security models are inadequate for handling the dynamics, scattered, and resource-limited characteristics of IoT-enabled educational systems. As a solution to these problems, this paper introduces ZeroTrustEdu, a scalable zero-trust cryptographic solution that combines lightweight post-quantum key management with adaptive trust scoring of cloud-connected IoT e-learning infrastructure. The proposed framework makes three fundamental contributions namely: (1) a hierarchical zero-trust security model with no implicit trust, operating across device, edge, and cloud layers; (2) a lightweight key distribution protocol based on the Module-Lattice Key Encapsulation Mechanism (ML-KEM) compliant with NIST FIPS 203 standards and (3) an adaptive behavioral trust scoring engine that dynamically adjusts device and user trust levels based on real-time interaction analytics. The architecture is evaluated using extensive NS-3 network simulations with up to 100,000 concurrent IoT nodes with formal security analysis under Chosen Plaintext Attack (CPA) and Chosen Ciphertext Attack (CCA) threat models. Comparative evaluation against RSA-2048, ECC-P256, and AES-256 baselines demonstrates that, ZeroTrustEdu delivers a 62% ± 3% (95% CI, 10 independent runs) reduction in ML-KEM encapsulation latency (12.8 ms for key encapsulation/decapsulation, contributing to a complete device authentication latency of 47.3 ms including ML-DSA signature operations), 45% reduced communication overheads, and 38% reduction in energy consumption on ARM Cortex-M4 constrained devices compared to RSA-2048 and achieves provable post-quantum security reducible to the hardness of the Module Learning With Errors (MLWE) problem. These findings demonstrate that the proposed architecture provides a viable, scalable, and quantum-resilient security solution for next-generation IoT-enabled e-learning environments. The cryptographic security of ZeroTrustEdu is guaranteed at the primitive level through NIST-standardized ML-KEM (FIPS 203) and ML-DSA (FIPS 204), with IND-CCA2 and EUF-CMA security formally proven in the respective standards; full protocol-level formal verification using automated theorem provers (ProVerif, Tamarin) is identified as valuable future work to rule out protocol-composition vulnerabilities beyond primitive-level guarantees. Full article

Background/Objectives: Lipidomics has become a key component of systems biology, enabling comprehensive characterisation of lipid species and their roles in health and disease. As regulators of membrane architecture, energy balance, inflammation, and cellular signalling, lipids offer a powerful framework for understanding metabolic dysfunction. In veterinary medicine, however, lipidomics remains comparatively underdeveloped. In cats, lipid metabolism is central to disorders such as hepatic lipidosis, cystitis, obesity, diabetes mellitus, and chronic inflammatory enteropathies, yet available data remain limited. This systematic review synthesised current evidence on lipidomics and lipid-focused profiling in feline disease and identified lipid alterations with potential clinical relevance. Methods: Following PRISMA 2020 guidelines, PubMed, ScienceDirect, and Scopus were searched for original studies (1994–2026) evaluating lipidomics or lipid-focused profiling in cats. Eligible studies assessed lipid species, fatty acids, lipid mediators, or lipoproteins in disease or physiological states. Owing to methodological heterogeneity, findings were synthesised narratively. Results: Seventeen studies met inclusion criteria, covering hepatic, urinary, gastrointestinal, renal, neurological, oncological, metabolic, and pharmacologically modulated conditions. Recurring alterations involved lipoproteins, triglycerides, phospholipids, sphingolipids, fatty acids, and oxylipins. More consistent patterns emerged in hepatic lipidosis, where lipoprotein disturbances may aid diagnosis; in lower urinary tract disease, where PUFA-derived oxylipins differentiated bacterial from idiopathic cystitis; and in obesity, where phospholipid and triglyceride shifts reflected metabolic risk. Fatty acid remodelling in chronic enteropathies aligned with mucosal inflammation, while sphingolipid changes in neurological disease correlated with severity. Heterogeneity in analytical platforms, dietary control, and study design limited comparability. Conclusions: Feline lipidomics reveals biologically meaningful alterations with emerging diagnostic and prognostic value. Although still developing, lipid-focused approaches may enhance disease characterisation and support translational research. Larger, standardised studies and robust reference datasets are needed to validate lipid signatures for clinical implementation. Full article

Severe mental disorders, including schizophrenia (SCZ), bipolar disorder (BD), and major depressive disorder (MDD), are leading causes of global disability, yet current treatments remain largely symptomatic and fail to alter disease trajectories. Converging evidence from genetics, longitudinal studies, and systems neuroscience supports a dimensional and transdiagnostic architecture of psychopathology, involving shared polygenic risk and overlapping neurodevelopmental and circuit-level alterations. Traditional approaches—such as post-mortem brain analysis, neuroimaging, and animal models—have delineated core molecular perturbations (e.g., dopaminergic, glutamatergic, and GABAergic dysfunction), as well as informed translational frameworks for mechanistic investigation, but remain constrained by restricted access to dynamic processes and incomplete recapitulation of human-specific biology. The advent of human-derived cellular models, particularly human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs), has partially addressed these limitations, enabling the study of patient-specific neurodevelopment and synaptic function in vitro. Within this evolving landscape, the olfactory neuroepithelium (ONE) has emerged as an accessible source of neural progenitors, obtainable through minimally invasive procedures, providing a window into living human neurobiology. ONE-derived cells retain donor-specific genetic and epigenetic signatures while recapitulating disease-relevant phenotypes across major psychiatric disorders, including altered neurodevelopmental dynamics, synaptic gene expression, and inflammatory profiles. Here, we present a narrative review of the principal cellular and tissue models used in biological psychiatry, examining their respective strengths, limitations, and translational relevance across experimental contexts. By situating these approaches within a unified framework, we aim to clarify their complementarity, identify current gaps, and outline future directions, highlighting the emerging potential of ONE-based models to bridge genetic risk, cellular dysfunction, and clinical phenotype, thereby advancing precision psychiatry. Full article

While a conserved core microbiome is shared across healthy individuals, significant interindividual taxonomic variation exists; however, the specific influence of genetic ancestry on supragingival plaque structure in eubiosis remains unclear. This systematic review analyzed evidence regarding taxonomic variations in supragingival plaque associated with ethnicity in systemically healthy populations. A search was conducted in PubMed, Scopus, ScienceDirect, and Scielo following PRISMA 2020 guidelines, covering literature up to October 2025. Cross-sectional studies using genomic sequencing or metagenomics were included, with quality assessed via the GRADE system. Six studies met eligibility criteria. Results identified a universal core microbiome structurally dominated by Corynebacterium spp. and Streptococcus spp. However, distinct ethnic-specific taxonomic signatures emerged, such as the enrichment of Fusobacterium spp. in African Americans and Corynebacterium spp. in Caucasians, alongside the exclusive presence of Sneathia spp. in Burmese individuals. Although a basal microbial architecture necessary for homeostasis exists, ethnicity acts as a biological filter defining distinctive bacterial profiles and differential susceptibilities. These findings suggest that while the core microbiome is conserved, the composition of peripheral species in the dental plaque hedgehog structure varies according to ancestry. This supports a transition from standardized dental care to personalized medicine oriented towards the patient’s biological heritage. Full article

Regular time-based preventive maintenance for medical devices often fails to detect actual component degradation. This study proposes a K-NN predictive framework that analyzes the residual current signal ($I_{\Delta }$) to categorize the operational conditions of medical devices across two representative device types: Syringe Pump and Patient Monitor. The raw signal was transformed into a higher-dimensional feature space, consisting of mean, standard deviation, gap, and RMS, to handle its characteristics. By evaluating various distance metrics, the results show that Cosine provides the most efficient diagnostic path, achieving optimal factor ($f_{opt}$) at a lower number of neighbor parameters, at K = 4 for Syringe Pump and K = 8 for Patient Monitor with an accuracy of 94.21% and 94.41%, respectively. The disparity in K-values reflects the inherent model complexity resulting from distinct power supply architectures, a characteristic also manifested in reactive power (Q). By mapping this statistical transformation, the model overcomes the limitations of static threshold-based leakage current monitoring. This research marks a paradigm shift towards a data-driven Energy-Centered Maintenance (ECM) strategy. By facilitating interventions triggered by empirically assessed signal signatures rather than predetermined time intervals, this framework optimizes maintenance activities and enhances the overall energy efficiency of hospital infrastructure. Full article

The polymer–filler interphase in filled elastomers is often represented by a single thickness, obscuring internal heterogeneity. Coupling coarse-grained molecular dynamics with dynamic mechanical analysis of EPDM/carbon-black compounds, we resolve a bimodal bound-rubber layer with a dense inner zone set by surface adsorption and a looser outer zone sustained by chain connectivity. Heating contracts the outer zone about twice as strongly as the inner zone (outer: 26.5%, 95% confidence interval 17.4–34.8%; inner: 13.3%). Per-layer mean-squared displacement analysis shows a modest mobility gradient between the 1–2 nm outer zone and the bulk. Dynamic mechanical analysis at 120–140 °C shows a flatter reinforcement factor at higher temperature, consistent with interphase-linked thermal contraction. Lengthening the chain at fixed filler loading markedly enlarges the bridging fraction and the cumulative excess thickness, signaling a transition from adsorption-limited to connectivity-limited reinforcement. These results show that a single interphase boundary can miss a dynamically active outer zone relevant to reinforcement and thermal aging in filled elastomers. Full article