Search Results (9,247)

The rapid advancement of large language models (LLMs) offers new avenues for overcoming the limitations of current data-driven GIS analysis methods, giving rise to the emerging field of large language model-driven GIS analysis (LLM4GIS). This study presents a systematic review of 673 relevant research articles published between 2022 and July 2026. By tracing the five-stage evolutionary trajectory of GIS analysis methods and elaborating the framework for LLM4GIS, we conduct a quantitative bibliometric analysis that reveals the field’s rapid growth from sporadic exploration to explosive expansion, along with the shifting landscape of its dominant technical trajectories. Building on these analyses, we develop a taxonomy spanning four dimensions—technical approaches, datasets, applications, and evaluation—and systematically analyze the advances and evaluation methodologies across five core application areas: geospatial knowledge question answering, knowledge extraction, spatiotemporal reasoning, geospatial code generation, and analysis modeling. Key challenges and future research directions are further discussed along five dimensions: multimodal data integration, high-quality data development, domain-specific model development, integration of world models with real-world geospatial environments, and interpretability and trustworthiness. This review provides a systematic reference for advancing the bidirectional synergy between LLM4GIS and GIS4LLM and for fostering a new generation of spatiotemporal intelligence ecosystems. Full article

Mycotoxigenic fungi contaminate cereals, nuts, fruits, and other food commodities, threatening human and animal health. This review focuses on Aspergillus flavus, Fusarium verticillioides, and Alternaria alternata as representative mycotoxigenic fungi because of their importance as food contaminants and their production of major food-relevant mycotoxins. Some methods detect fungal presence or toxigenic potential, whereas others directly quantify toxins. Fungal identification or detection of biosynthetic genes indicates presence or toxin-producing potential, respectively, but neither replaces direct mycotoxin measurement. The diversity of targets, toxins, and food matrices remains a challenge for accurate and practical analysis. Particular attention is given to immunodiagnostic approaches, including enzyme-linked immunosorbent assays (ELISAs), as well as complementary formats such as lateral flow immunoassays (LFIAs), and immunosensors, for laboratory and on-site food safety monitoring. The advantages and limitations of these approaches, with emphasis on sensitivity, specificity, matrix effects, and sample preparation, are discussed. Future aspects will focus on integrated, matrix-validated workflows combining robust sample preparation, multiplex recognition, portable readout, and appropriate quality controls to enable faster and more reliable monitoring of fungal contamination and mycotoxins throughout the food and feed chain. Full article

Background: Delayed discharge contributes to prolonged hospital stays, particularly in inpatient rehabilitation. This study explored barriers and facilitators to implementing the Clinical Frailty Scale (CFS) and a frailty intervention bundle in a rehabilitation hospital. Methods: We conducted a qualitative descriptive study using five semi-structured focus groups with 27 clinicians and organizational leaders across three inpatient rehabilitation units at Hennick Bridgepoint Hospital in Toronto, Canada, in July and August 2025. We analyzed data using deductive, directed content analysis informed by the Consolidated Framework for Implementation Research. Results: Facilitators included iterative adaptation, alignment with provincial priorities, training, early staff engagement, leadership support, and ongoing feedback. Barriers included limited perceived added value, duplicate documentation, uncertainty about rehabilitation evidence, community resource shortages, change fatigue, inconsistent information sharing, and weak feedback linking implementation to patient outcomes. Participants viewed the CFS as confirming clinical judgment while providing standardized communication and a shared understanding of risk across care levels. Conclusions: Participants identified contextual fit, workflow integration, and responsiveness to frontline feedback as important facilitators of early implementation. Adaptable, context-sensitive approaches may strengthen future implementation and sustainability of frailty screening and intervention bundles in post-acute rehabilitation. Full article

This study examines the development and intellectual structure of artificial intelligence (AI)-based financial fraud detection through a bibliometric analysis of 2200 Scopus- and Web of Science-indexed journal articles published between 1992 and 2026. The literature has expanded rapidly, with 89.6% of publications appearing since 2018 and an annual growth rate of 29.4%. Research is concentrated in Computer Science and Engineering, whereas Business and Economics account for 9.2% of classified publications. Credit card and payment fraud constitutes the largest research domain (32.0%), followed by banking, loan, and transaction fraud (14.3%), insurance fraud (12.2%), and accounting and financial statement fraud (9.9%). Neural networks and deep learning, ensemble and boosting methods, and classical supervised classifiers are the dominant methodological families. The field has progressively shifted from traditional data-mining and classical machine-learning approaches toward deep learning, ensemble methods, graph neural networks, and explainable AI. Science mapping shows that the intellectual structure remains strongly centered on payment fraud, while accounting and insurance fraud form more distinct research fronts. Author productivity is highly dispersed, with 88.9% of authors contributing a single publication, while international collaboration varies considerably across leading countries. By integrating performance analysis, science mapping, fraud-domain classification, bibliometric laws, and publication-growth analysis, the study provides a comprehensive assessment of the field’s development, intellectual structure, and emerging research directions. Full article

Background and Objectives: Elevated intracranial pressure (ICP) is common to many encephalopathies, but the evolution of quantitative electroencephalography (qEEG) indices along a pressure gradient—and their relative suitability for early warning—have not been systematically compared. In this exploratory study we established a stepwise, steady-state rabbit model of graded intracranial hypertension and screened candidate early-warning features against prespecified criteria. Materials and Methods: Eight male New Zealand rabbits were studied; two were excluded—one for circulatory failure during catheterization and one for refractory status epilepticus that reached a humane endpoint—leaving six that completed the protocol. Repeated small-volume inflation of a left parenchymal water balloon was used to define five windows: baseline (BL; ICP < 10 mmHg), 15–20, 25–30 and 35–40 mmHg, and an end stage (END). Eight-channel EEG was recorded synchronously, and δ, θ, α and β power, together with the amplitude-integrated EEG (aEEG) upper margin, lower margin and bandwidth, were computed. Seven candidate features and three criteria (evolution timing, individual consistency and strength of association with ICP) were fixed before analysis; selection stability was assessed by leave-one-rabbit-out (LORO) cross-validation. Results: ICP rose in a controlled manner from 5.43 ± 1.09 to 35.93 ± 1.48 mmHg (Kendall W = 0.96), without systematic changes in respiratory mechanics. As ICP increased, absolute θ and α power declined progressively (ρ = −0.77 and −0.53), whereas absolute δ power did not increase and its rising relative share merely reflected redistribution after the loss of faster activity. The three aEEG components changed asynchronously: the lower margin shifted downward early and persistently (ρ = −0.87 with ICP; W = 0.86), whereas the upper margin changed only terminally and bandwidth showed no consistent direction. Concurrently, MAP rose through the Cushing response, with heart-rate slowing, but the rise was insufficient and CPP fell from 64.78 to 41.24 mmHg, approaching the rabbit lower limit of autoregulation (~50 mmHg CPP) only in the later windows. Because ICP and CPP were tightly coupled within animals, their independent effects on the EEG could not be separated. In leave-one-rabbit-out cross-validation, the lower margin was selected in five of six folds, with a median out-of-sample ρ of −1.00 (a discrete value based on four windows per animal). Conclusions: In this exploratory study in a small cohort (n = 6), qEEG indices evolved asynchronously along the ICP gradient; the aEEG lower margin appeared earliest and remained relatively stable across leave-one-rabbit-out folds, making it a candidate early-warning feature worthy of priority validation. However, because rising ICP and falling CPP were tightly coupled and could not be separated in this model, its early decline cannot be uniquely attributed to either ICP elevation or hypoperfusion. Its alarm threshold and early-warning value still require confirmation in larger independent cohorts with simultaneous CPP and cerebral blood flow monitoring. Full article

Deep learning for biomedical signal analysis is revolutionizing the analysis of complex physiological signals through the extraction of intricate patterns from physiological and wearable sensor data. This Special Issue brings together five articles that address cardiac, neural, muscular, and motion signals using complementary approaches across signal processing, machine learning, deep representation learning, and wearable sensing. Two contributions focus on ECG: one introduces a constraint-based method to suppress false positives in P/R/T peak detection, and the other one presents an interactive GUI for rapid R-peak correction. In the third study, an EEG biometrics system was developed that uses 3D brain mapping and autoencoder features, and the fourth study compares machine classifiers for athlete movement recognition from EMG. In the final study, wearable IMUs were used to improve jump height estimation via optimal sensor placement. We summarize these works and highlight themes such as physiological prior integration, human–AI interaction, and multi-sensor fusion. This Editorial concludes with the identification of future directions, including multi-modal modeling and adaptive learning in real-world settings. Full article

This review examines current approaches to the development of multicomponent functional systems based on secondary aluminum alloys, mechanochemically modified metal powders, metal–organic frameworks, and aerosol-generating materials. It is shown that controlling the morphology, particle size, phase composition, and surface state of aluminum-containing powders is among the key factors governing their reactivity in high-temperature processes. Particular attention is paid to mechanochemical activation as a method for the controlled modification of particle structure, enhancement of crystal lattice defect density, formation of new interfacial contacts, and surface modification. The potential of metal–organic frameworks for interfacial modification, structural functionalization, and incorporation of active sites is critically assessed, while emphasizing that direct experimental evidence for their integration with secondary-aluminum-based powder systems remains limited. The use of carbon-containing and mineral materials for the preparation of aerosol-generating systems with tailored optical and spectral characteristics is analyzed separately. The application of secondary aluminum alloys is shown to provide opportunities for reducing production costs, improving resource efficiency, and advancing import-substitution technologies. The review identifies the integration of secondary aluminum alloys, mechanochemical activation, SHS, and MOF-based functionalization as an emerging research direction and highlights the need for further experimental validation, quantitative comparison, and pilot-scale assessment before such systems can be regarded as an established technological platform. Full article

Traditional consumer credit scoring models frequently fail to identify rising default risks in transition economies because they rely almost exclusively on formal macroeconomic indicators and unchanging demographic data. This paper offers a preliminary empirical assessment of digital gambling participation and utility payment arrears as stand-in risk indicators in Kazakhstani consumer credit scoring. Using a unique survey-derived dataset from Kazakhstan (N = 41 survey respondents, yielding a finalized modeling cohort of N = 233 active borrowers), we develop, tune, and evaluate five predictive models—Logistic Regression (LR), Decision Trees, Artificial Neural Networks (ANNs), Random Forests, and Extreme Gradient Boosting (XGBoost). Models were optimized using stratified 5-fold cross-validation on the training partition (N ≈ 163) and benchmarked across both a SMOTE-balanced evaluation pool (D_eval, N = 473) and an empirical 30% holdout test set (D_emp, N ≈ 70). The results provide localized baseline evidence on incorporating non-traditional behavioral data into credit evaluation frameworks in data-constrained emerging economies. While the raw empirical cohort (𝒟_emp) exhibits a dominant majority distribution with an 88.84% default/late-payment rate, a balanced resampled evaluation pool (𝒟_eval, N = 473) was constructed to ensure unbiased model optimization. The empirical results show that tree-based ensemble structures considerably outperform conventional methods, establishing a clear performance hierarchy. With an accuracy of 83.1% and an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.854, the classical LR baseline has the lowest predictive power and is unable to map complex, interactive behavioral hazards. On the other hand, the best architecture is the improved XGBoost ensemble engine, which achieves an AUROC of 0.976 on the balanced evaluation pool (𝒟_eval). The XGBoost framework successfully reduces institutional asymmetric loss constraints, according to a micro-structural audit of the underlying confusion matrix. Evaluated on 𝒟_eval, the model achieves a specificity of 97.4% (TN = 228) and a sensitivity of 96.2% (TP = 230), reducing catastrophic FNR underwriting errors while preserving high loan onboarding efficiency. Furthermore, comparative testing directly on the unadjusted empirical cohort (𝒟_emp) confirms strong diagnostic robustness, demonstrating that high class-separation performance generalizes effectively to skewed real-world credit distributions. These findings demonstrate that incorporating real-time behavioral abnormalities into regularized ensemble pipelines offers a very reliable method of reducing risk. In conclusion, this research provides emerging market financial institutions and regulators with an empirical guide to enhance credit underwriting accuracy, strike a compromise between inclusive lending and capital preservation, and shield the retail banking portfolio from localized default contagion. Although the size of the available dataset does not allow us to attain a definitive conclusion, these results point to a promising direction to further investigate with the same methods proposed in this work and larger datasets. Indeed, they suggest that behavioral data has the potential to enhance traditional credit-scoring techniques when properly vetted and integrated with machine-learning techniques. Full article

The Monte Carlo (MC) method is the reference approach for particle transport simulation in complex three-dimensional geometries and continuous-energy problems, where it avoids the phase-space discretization required by deterministic methods; yet its $1/\sqrt{N}$ convergence rate renders brute-force computation infeasible for deep-penetration shielding and full-core fixed-source problems. Variance reduction (VR) methods address this bottleneck by modifying particle sampling or transport while preserving estimator unbiasedness, concentrating computational resources on high-contribution phase-space regions. This review systematically surveys VR methods for Monte Carlo particle transport, organized into three categories: classical techniques, advanced local variance reduction (LVR), and advanced global variance reduction (GVR). Using Zero-Variance Theory as the unifying framework, the review elucidates that all VR methods are fundamentally distinct approximations to the optimal importance function. A multi-dimensional classification system is proposed that traces the evolution from manual, heuristic parameter tuning to automated generation, from single-technique application to multi-solver collaborative integration, and from empirical rules to mathematically rigorous efficiency optimization. Key future directions include the rigorization of global efficiency theory, machine-learning-assisted importance function construction, and multi-solver collaborative adaptive frameworks. The field is progressing toward a hierarchical complementarity of local and global VR, jointly advancing VR methodology from academic research to engineering practice. Full article

Electrohysterography (EHG) is an advanced, noninvasive technique for characterizing myometrial activity from the maternal abdominal surface. Despite its clinical relevance, the adoption of EHG in wearable systems is hindered by variability in electrode configurations and a lack of integrated system design. A PRISMA 2020-based systematic review of publications from September 1997 to January 2026 was conducted using IEEE Xplore, PubMed, Semantic Scholar, ScienceDirect, SpringerLink, MDPI, IOP, and Frontiers, supplemented by Crossref searches via Publish or Perish and manual citation tracking. Of the 1084 records identified, 91 publications were included in the evidence synthesis. The analysis examined electrode configurations, wearable design, and measurement methodologies, focusing on the effects of electrode density and placement on signal quality and clinical usability. The reviewed literature indicates that configurations employing approximately 8–16 electrodes represent a practical range for balancing spatial characterization and clinical usability. However, the findings reveal persistent challenges related to the limited integration of electrode design, signal acquisition, and analytical methods. This evidence underscores the need for an integrated approach that considers flexible electrode interfaces, standardized electrode configurations, and adaptive signal processing techniques. By aligning technological improvements with clinical requirements, this study identifies practical considerations for developing wearable maternal–fetal monitoring systems. Full article

The dense control networks constructed for regional and global image blocks often contain a substantial number of redundant tie points that increase the computational burden of bundle adjustment (BA). Reducing this redundancy without weakening the geometric support of image pairs is essential for subsequent photogrammetric processing. Existing image-centered network-thinning methods reduce point counts but do not explicitly preserve direct links between individual image pairs. This paper presents an efficient control-network thinning method for planetary photogrammetry based on image-pair convex-hull computation and grid-based interval sampling. The proposed method first organizes the input control network into an image-pair measure map. It then separates each pair’s measures into hull and interior classes using Andrew’s monotone-chain algorithm and samples the two classes at different intervals. When a tie point is selected for an image pair, its measures in both images are retained together. An optional rule can also retain measures of the same surface feature in other overlapping images to help maintain connections across the image block. The method is designed to thin an existing control network after upstream quality filtering. It does not re-evaluate tie-point quality or add new control points. We implemented the method as the cnetoptimize application within the Integrated Software for Imagers and Spectrometers (ISIS), a widely used open-source planetary photogrammetric software system developed by the United States Geological Survey (USGS). Experiments used a large-scale Lunar Reconnaissance Orbiter (LRO) Narrow Angle Camera (NAC) network and a medium-scale Mars Express (MEX) High Resolution Stereo Camera (HRSC) network. At comparable compression ratios, cnetoptimize retained strict non-degenerate support for more image pairs than the image-centered ISIS cnetthinner. Across the tested datasets and settings, cnetoptimize preserved the direct connections of all originally non-degenerate image pairs at compression ratios up to 8.4×. In a downstream test on one weak HRSC pair, bundle adjustment with the thinned cnetoptimize network yielded subpixel image residuals and a local 25 m digital orthophoto map (DOM) seam root mean square (RMS) of 36.0 m. The method provides an ISIS-compatible approach to control-network thinning for planetary mapping. Full article

Knowledge tracing (KT) models the evolution of students’ knowledge states from historical learning interactions, serving as a fundamental mechanism for intelligent educational systems. The efficacy of KT models heavily relies on the quality of question representations. However, existing methods incorporating concept relations generally face two critical limitations: First, static relation graphs fail to capture dynamic cognitive trajectories, while shallow statistical approaches are highly vulnerable to sparse data and interaction noise. Second, the current models lack fine-grained semantic characterization, specifically failing to integrate concept difficulty to accurately evaluate the cognitive load of individual questions. To address these issues, we propose SSCKT, a novel framework that adaptively extracts latent concept relations and introduces concept associations to enhance question representation via structural–semantic co-optimization. Specifically, a concept relation discovery and aggregation module dynamically learns a directed acyclic graph (DAG) of concept relations from historical learning interactions, which is encoded via heterogeneous graph convolutions to yield structure-enhanced embeddings. Concurrently, a concept difficulty-guided semantic enhancement module utilizes multi-head attention to capture intrinsic question–concept dependencies. This module employs a topological concept difficulty metric as a scaling factor to adjust the question offset vector, enabling cognitive load-aware modeling. Finally, a contrastive learning mechanism aligns the structural and semantic representations to eliminate feature redundancy, generating highly discriminative question embeddings. Extensive experiments on five public datasets demonstrate that, relative to the strongest baseline on each dataset, SSCKT improves average ACC and AUC by 1.20% and 1.33%, respectively, validating the superiority of the proposed co-optimization architecture. Full article

Objective: To evaluate the incremental predictive value of pretreatment tri-modal ultrasound deep learning integrating two-dimensional ultrasound (2DUS), color Doppler flow imaging (CDFI), and contrast-enhanced ultrasound (CEUS) for predicting breast pathologic complete response (bpCR) after neoadjuvant chemotherapy (NAC), and to assess performance stability across independent centers. Methods: This retrospective multicenter study included 1089 patients with primary invasive breast cancer. Center 1 provided the training cohort (n = 572) and internal validation cohort (n = 246), whereas centers 2 (n = 160) and 3 (n = 111) served as independent external validation cohorts. Using pretreatment 2DUS, CDFI, and CEUS within a standardized intratumoral and 10-mm peritumoral framework, single-modality, all three dual-modality combinations, and tri-modal models were developed. The best-performing dual-modality model was identified and locked during training as the primary imaging comparator. A Tri-modal + Clinical model further integrated pretreatment clinical and biomarker variables. Performance was assessed using AUC, Brier score, calibration, decision curve analysis, and ablation experiments. Results: Overall, 383 patients (35.2%) achieved bpCR. CEUS was the strongest single modality, and 2DUS + CEUS was the best-performing dual-modality model. Compared with 2DUS + CEUS, the tri-modal model increased the AUC from 0.895 to 0.919 in the combined external cohort (ΔAUC, +0.023; 95% CI, +0.010 to +0.038; p = 0.001), with corresponding improvements in Brier score (0.124 to 0.105) and F1 score (0.747 to 0.786). The AUC point estimate increased in both external centers, although the 95% CI for the incremental AUC included zero in external validation cohort 2. Despite modest standalone discrimination, CDFI provided measurable incremental information when added to 2DUS + CEUS. Further integration of pretreatment clinical and biomarker variables increased the combined external AUC to 0.935. Reduced-input retraining showed performance decreases after omission of any modality, with the largest reduction after CEUS omission. Conclusions: Pretreatment tri-modal ultrasound deep learning provided incremental predictive value for bpCR beyond the best dual-modality model, with a consistent direction of benefit across two independent external centers. Clinical and biomarker integration provided additional predictive information, supporting further evaluation of this multimodal framework for individualized pretreatment bpCR prediction. Full article

The increasing integration of distributed energy resources (DERs), particularly photovoltaic (PV) systems, is transforming modern power systems and introducing new challenges for protection, reliability, and stability. The transition toward smart grids with bidirectional power flows has significantly changed the operating conditions under which protection systems must perform. Among the key challenges associated with DER integration is unintentional islanding, in which a DER or a group of DERs continues to energize a portion of the distribution network after disconnection from the main utility grid. If not detected and cleared within the required time, islanding may pose risks to personnel safety, power quality, equipment, and system stability. This paper presents a comprehensive review and critical analysis of islanding detection methods for PV and inverter-based DER systems, including passive, active, hybrid, and communication-based techniques. The methods are evaluated with respect to detection speed, the non-detection zone (NDZ), false-trip behavior, power-quality impact, implementation complexity, cost, and adaptability to different inverter and grid conditions. Particular attention is given to multi-inverter systems, weak and converter-dominated grids, communication and cybersecurity, and standardization and benchmarking challenges. Based on the identified limitations and research gaps, future research directions are discussed, with emphasis on AI/ML-based detection, adaptive protection, grid-forming inverters, multi-inverter coordination, PV–battery systems, weak-grid and low-inertia networks, and experimental validation. The review highlights the need for adaptive, coordinated, and intelligent islanding protection frameworks capable of ensuring reliable detection in increasingly complex inverter-dominated power systems. Full article

Background/Objectives: The vertebral column is exposed to significant mechanical loads, exceeding 8000 N at certain levels due to bipedal posture and locomotion. These forces may contribute to structural alterations during growth. Scoliosis is among the most common structural spinal conditions in children and adolescents, with reported prevalence ranging from 0.5% to 5.2%. Our study aims to evaluate the radiographic characteristics of scoliosis in pediatric, adolescent and young adult patients and to analyze associations between curve severity and anatomical and developmental parameters. Methods: This retrospective cross-sectional study included 233 patients (152 females, 81 males) examined between August and October 2023. Cobb angle, curve localization, curve direction, Risser stage, and spina bifida occulta were assessed. Associations between variables were analyzed using chi-square and Fisher’s exact tests, and multivariable linear and ordinal regression models. Results: Scoliosis (Cobb angle ≥ 10°) was identified in 62.2% of patients within this referral cohort. A statistically significant association was observed between curve localization and scoliosis severity (p = 0.007), with thoracic curves being present more frequently with higher Cobb angles. No statistically significant associations were found between scoliosis severity and sex or curve direction. Spina bifida occulta was detected in 11.7% of scoliotic patients. Skeletal maturity followed expected age-related patterns. In multivariable ordinal logistic regression across the entire cohort, primary thoracic curve localization (aOR = 4.64, p < 0.001), female sex (aOR = 3.06, p < 0.001), and older age (aOR = 1.18, p = 0.002) were independently associated with higher scoliosis severity categories, whereas Risser stage and spina bifida occulta were not. In subgroup multivariable linear regression restricted to thoracic curves, higher Risser stage (β = −3.11°/stage, p = 0.041) and older age (β = 2.12°/year, p = 0.013) were independently associated with continuous Cobb angle magnitude. Conclusions: Scoliosis severity is associated with anatomical curve localization and skeletal maturity demonstrates age-appropriate ossification patterns. These findings support the value of integrated radiographic assessment, including Cobb angle evaluation and Risser staging, in the diagnostic work-up of pediatric, adolescent, and young adult scoliosis. Full article