Search Results (13,818)

In complex and unknown processes, global models are fitted over the entire input domain but often tend to perform poorly whenever the response surface exhibits non-stationary behavior and varying smoothness. A common approach is to use local models, which requires partitioning the input domain into subdomains and training multiple models, thereby adding significant complexity. Recognizing this limitation, this study addresses the need for models that represent the input–output relationship consistently over the full domain while still adapting to local variations in the response. It introduces a novel machine learning approach: the Polynomial Chaos Expanded Gaussian Process (PCEGP), leveraging polynomial chaos expansion to calculate input-dependent hyperparameters of the Gaussian process (GP). This provides a mathematically interpretable approach that incorporates non-stationary covariance functions and heteroscedastic noise estimation to generate locally adapted models. The model performance is compared to different algorithms in benchmark tests for regression tasks. The results demonstrate low prediction errors of the PCEGP, highlighting model performance that is often competitive with or better than previous methods. A key advantage of the presented model is its interpretable hyperparameters along with training and prediction runtimes comparable to those of a standard GP. Full article

The utilisation of composite materials has the potential to play a vital role in the development of lightweight structures for future generations of aircraft, with the objective to reduce emissions. Ultrasonic welding is a process that has been proven to exhibit advantageous qualities, including the capacity to achieve welds with a comparatively short process time. Furthermore, its capacity to function as both a static and a continuous process makes it a viable candidate for facilitating the realisation of this objective. The present study investigates the potential of a novel explicit modelling approach for the static ultrasonic welding process to more accurately represent the welding process by incorporating a more precise representation of the hammering effect. The hammering effect describes the partial loss of contact between the sonotrode and the upper adherend. The model’s validation was achieved through a multifaceted approach that incorporates high-speed camera recording, encompassing digital image correlation, laser displacement sensor measurements, and static ultrasonic welding experiments. These experiments encompassed varying welding times, followed by fracture surface analysis. The findings showed that an explicit time-domain model can effectively represent the static welding process of unidirectional materials utilising a film energy director. The experimental validation demonstrated a high degree of correlation between the thermal behaviour of the welding interface and the simulation results. The study demonstrated that the neutral position of the sonotrode exhibited an increase during the initial phase of the welding process due to dynamic stresses. This phenomenon enables reduced constraint movement of the adherends and the energy director, which results in the disconnection of the sonotrode from both the upper adherend and the energy director, as well as the adherends and the anvil. The higher neutral position of the sonotrode was then implemented in an explicit simulation of the static ultrasonic welding process. Full article

Over the past few decades, electrosynthesis has advanced significantly, enabling numerous valuable transformations for synthetic chemists. Olefins are inexpensive, readily available industrial feedstocks extensively used in organic synthesis. Therefore, achieving high-value transformation of olefins is of great value. However, the use of stoichiometric oxidants and the generation of stoichiometric waste hinder its broader application. Utilizing electrochemistry to achieve high-value transformations of olefins represents a green, environmentally friendly, and sustainable strategy, since it eliminates the need for external oxidants. This review discusses recent advances in the high-value conversion of alkenes induced by electrochemistry. The article introduces two modes of electrochemical olefin transformation, discussing both synthetic applications and mechanistic studies. It highlights their advantages and suggests future directions to tackle the existing challenges in this synthetic domain. Full article

This study presents a systematic comparative benchmark between two distinct paradigms for solving nonlinear partial differential equations (PDEs): the data-driven Physics-Informed Neural Networks (PINNs) and the analytical Homotopy Analysis Method (HAM). We apply both methods to a unified family of canonical PDEs, the Burgers, Huxley, Fisher, Burgers–Huxley, and Burgers–Fisher equations, under identical problem setups, domain discretization, and validation metrics. PINNs incorporate physical laws directly into neural network training by minimizing a loss function that enforces PDE residuals, yielding physically consistent solutions even for strongly nonlinear problems. HAM provides approximate analytical solutions using a unified framework, and the same initial guess, auxiliary linear operator, and auxiliary function across all equations despite their distinct nonlinearities. The controlled, consistent application of both methods enables a fair, reproducible comparison across this equation family. The results provide a quantitative performance map under identical conditions, delineating when PINNs (high accuracy, long-term stability, and generalization capability) are preferable, versus when HAM (computational speed, short-term analytic approximation, and lower memory footprint) offers advantages. While the finite radius of convergence of the truncated HAM series is theoretically expected, our controlled comparison quantifies for the first time how this degradation varies across equation types, revealing that the choice between methods depends on specific problem requirements including error tolerance, available computational resources, and temporal horizon. The novelty lies not in solving each equation individually, but in deriving a performance taxonomy that systematically connects equation features (shocks, stiffness, and reaction–diffusion coupling) to optimal solver choice—providing previously unavailable, evidence-based guidance for the scientific computing community. This study establishes the first rigorous, controlled comparative benchmark between analytic and data-driven PDE solvers across a spectrum of nonlinearities, providing a reproducible baseline for future hybrid scientific machine learning solvers. Full article

The development of antimicrobial resistance (AMR) and the emergence of multiresistant pathogens represent a growing global threat to both human and animal health. Beyond the excessive and improper use of antimicrobials in human medicine, irrational use in veterinary medicine, agriculture, and aquaculture significantly contributes to the selection and spread of resistant microorganisms, which can enter the food chain and reach humans through food consumption or handling. Based on results from a recent meta-analysis, the prevalence of antimicrobial-resistant foodborne pathogens in food samples exceeds 10%. The veterinary sector is of particular concern, as a large proportion of antimicrobials are used in animal production, generating strong selective pressure and favoring the dissemination of AMR along the food chain. In an increasingly interconnected global context, resistant pathogens and resistance determinants can disseminate rapidly across sectors and national borders, making strategies confined to a single sector insufficient; therefore, effectively addressing AMR requires a One Health approach encompassing the human, veterinary, and environmental domains. Key mitigation strategies include strengthening antimicrobial stewardship programs, also in animal production, reducing routine prophylactic use of antimicrobials, and improving surveillance, coordinated across sectors and, where possible, further supported by advanced technologies such as artificial intelligence and machine learning. Further efforts are also needed to improve microbiological diagnostics, particularly through rapid and molecular methods, to support timely, targeted therapies and reduce inappropriate empirical treatments. In parallel, investment in new therapeutic options, including innovative molecules, drug combinations, and alternative approaches, remains crucial to effectively countering the growing burden of antimicrobial resistance. Full article

Background: Encouraging fruit and vegetable (FV) consumption early in childhood is important for long-term healthy eating. Though parents play an important role in shaping children’s FV-related taste preferences and consumption, validated instruments assessing the range of parenting practices that specifically support young children’s FV intake are scarce. Furthermore, little attention has been given to low-income families, cultural inclusivity, and FV practices across different settings. The current study sought to conduct an initial examination and explore the measurement equivalency of a new FV parenting practices questionnaire (FVPPQ) across racially/ethnically diverse groups that address these gaps. Methods: Data for this paper came from a large project focused on parents’ FV parenting practices with young children enrolled in Head Start programs in the southern part of the U.S. Inclusion criteria were (a) parent/legal guardian being eighteen or older, (b) being the primary person responsible for child feeding, and (c) the child not requiring a special diet (e.g., diabetic). Using a multi-phases project approach, we (1) developed a preliminary conceptual map of parenting practice domains by reviewing existing measures on FV parenting practices; (2) completed and content-analyzed data from 18 focus groups (n = 62) to identify and further revise the preliminary conceptual map of domains, (3) administered a questionnaire with 11 domains of FV parenting practices, and then (4) empirically explored and reduced the measure while evaluating its content, construct, and criterion validity, and cultural equivalency across Non-Hispanic White, Hispanic White, and Black parents (n = 281). Results: Findings from Phases 1 and 2 generated a 107-item questionnaire that was reduced during phase 3 through a series of principal component and confirmatory factor analyses to the final FVPPQ with 21 items in four unique domains, showing good variability and inter-item consistency reliability: (1) Availability (5 items); (2) modeling (5 items); child-focused (5 items); and pressure (6 items). Three of the four domains evidenced cultural equivalency. Conclusions: The FVPPQ with four unique subscales demonstrated good content, construct validity, and partial measurement equivalency across racially/ethnically diverse groups of parents. Further confirmatory validation is warranted in larger samples, but the FVPPQ might be a promising and easily administered measure for research and applied interventions in nutrition, health behavior, and parenting contexts. Full article

With the growing significance of renewable energy resources and energy storage systems, the number of grid-connected inverters has been rising at an increasingly rapid pace. Generally, these inverters are directly integrated with the distribution network by synchronizing with the grid voltage at the point of common coupling. However, the low grid strength and varying R/X ratios, as the common characteristics of most distribution networks or weak grids, can lead to dynamic interactions that comprise stability and limit the power transfer capacity of grid-connected inverters. To ensure stable operation of the inverters, researchers must determine the stability boundary, described as the maximum power transfer capacity of grid-connected inverters under the premise of maintaining system small-signal stability. For this purpose, we propose to formulate a state-space model of the system in the synchronously rotating dq-frame of reference and perform eigenvalue analysis to determine the stability boundary. With a detailed model of the control structure and parameters of the grid-connected inverters, the stability boundary is identified as a surface with respect to different grid strengths and R/X ratios. Case study results of proposed eigenvalue analysis are compared with those of admittance model-based stability analysis as well as time-domain simulation using a switching model in Matlab/Simulink, validating the effectiveness and accuracy of the proposed eigenvalue analysis for stability boundary identification. Full article

We propose Trust Triangle, a Bridging Methodology that establishes evidential reliability through multi-attribution consensus, ensures external validity via statistical hypothesis testing, and enables controlled generation with RAG-anchored LLMs to transform black-box predictions into trustworthy, auditable explanations. This framework is instantiated for credit card fraud detection by integrating multi-method feature attributions with rigorous statistical validation. The resulting reliability-validity-verified insights are synthesized with high-relevance domain knowledge (relevance score > 0.7) retrieved from a real-world corpus via Retrieval-Augmented Generation (RAG). A structured Chain-of-Thought (CoT) prompt then guides an LLM to produce coherent, audit-ready case reports. Our contributions are threefold: (1) a verifiable framework for quantifying attribution reliability and validity, (2) a demonstrated end-to-end pipeline from robust prediction to semantically grounded explanation, and (3) a generalizable paradigm for Trustworthy ML in high-stakes domains. Experiments on a highly imbalanced dataset (fraud rate: 8.74%) demonstrate robust performance (PR-AUC = 0.7867), successfully identify statistically significant predictive features, and generate audit-ready reports, thereby advancing a rigorous, evidence-based pathway from model output to decision-ready support. Full article

Background/Objectives: Mild cognitive impairment (MCI) is common in Parkinson’s disease (PD) and significantly impacts quality of life. Non-invasive brain stimulation (NIBS) techniques have emerged as potential therapeutic interventions. This systematic review analyzes the current evidence regarding the efficacy of Transcranial magnetic stimulation (TMS) and transcranial electrical stimulation (tES) on cognitive domains in patients with PD-MCI. Methods: A systematic search was conducted across the PubMed, Scopus, Web of Science, and Medline Ultimate databases up to 20 November 2025. We included studies investigating the effects of NIBS compared to sham stimulation on neuropsychological outcomes in diagnosed PD-MCI patients. Results: Eight studies involving different stimulation protocols were included. Interventions primarily used TMS or tES targeting the left dorsolateral prefrontal cortex (DLPFC). Episodic memory and global cognition were the most responsive domains, assessed with specific neuropsychological scales. Findings for executive functions and attention were heterogeneous, while visuospatial abilities generally showed limited immediate response. Conclusions: NIBS represents a promising but low-certainty-evidence adjunctive therapy for PD-MCI, with improvements found in memory and global cognition. Future research should prioritize larger sample sizes, combined interventions (NIBS plus cognitive rehabilitation), and extended follow-ups to evaluate long-term neuroplasticity. Full article

Electroencephalography (EEG) has become an increasingly important tool in depression research due to its ability to capture objective neurophysiological abnormalities associated with depressive disorders, offering high temporal resolution, non-invasiveness, and cost-effectiveness.However, existing methods often fail to fully exploit the multi-domain information in EEG signals, resulting in limited model generalization capabilities. This paper proposes an improved bidirectional long short-term memory (BiLSTM) model that segments continuous EEG into non-overlapping 2-s epochs and learns end-to-end from multi-channel temporal sequences. After band-pass filtering and resampling, each epoch is represented as a channel–time matrix $X\in {\mathbb{R}}^{C\times T}$ (with C = 128) and processed by a BiLSTM encoder followed by a dynamic-routing encapsulated-vector classifier. On the MODMA dataset under subject-independent five-fold cross-validation, the proposed method outperforms a set of reproduced representative baselines (SVM, EEGNet, InceptionNet, Self-attention-CNN and CNN–LSTM) and achieves 84.8% accuracy with an AUC of 0.899. We further discuss recent contemporary directions (e.g., attention/Transformer-based and emotion-aware expert models) and clarify the scope of our empirical comparisons. Furthermore, experiments comparing different frequency bands and band combinations indicate that joint multi-frequency input can enhance classification performance. This study provides an effective multi-domain fusion approach for the automatic diagnosis of depression based on EEG. Full article

Background: Inherited kidney diseases represent a genetically and clinically heterogeneous group of disorders affecting both pediatric and adult populations. Advances in next-generation sequencing (NGS) have improved diagnostic precision; however, genotype–phenotype correlations and diagnostic yield vary substantially across disease entities. Methods:We retrospectively evaluated 165 patients referred for genetic testing due to suspected inherited kidney disease. Patients were classified into three clinical groups: polycystic kidney disease, Alport syndrome, and other syndromic patients with inherited kidney diseases. Genetic analysis was performed using NGS with Human Phenotype Ontology–based gene filtering and included evaluation of both single-nucleotide variants and copy number variations. Results: Overall diagnostic yield differed markedly between groups. A molecular diagnosis was achieved in 71.4% of Alport patients, 41.0% of PKD patients, and 70.2% of patients in the Other syndromic group. In the Alport group, variants were identified exclusively in COL4A3, COL4A4, and COL4A5, with pathogenicity and gene involvement correlating with disease severity and the presence of extrarenal manifestations. The PKD group showed predominant involvement of PKD1, followed by PKHD1 and PKD2, while a substantial proportion of patients remained genetically negative, reflecting technical and biological complexity. The Other group exhibited pronounced genetic heterogeneity, with variants distributed across multiple genes involved in tubular, glomerular, metabolic, and ciliopathy-related pathways. Computational assessments demonstrated that several variants of uncertain significance (VUS) were located in functionally critical domains and were predicted to disrupt protein stability, intermolecular interactions, or conserved structural motifs, thereby supporting the biological plausibility of their potential pathogenic impact. Conclusions: Phenotype-driven NGS enables effective molecular diagnosis across diverse inherited kidney diseases while revealing disease-specific differences in diagnostic yield and genotype–phenotype correlations. Systematic inclusion of variants of uncertain significance and careful integration of genetic and clinical data are essential for accurate interpretation and long-term patient management. Collectively, this study enhances understanding of inherited kidney diseases and underscores the value of integrating comprehensive genomic and computational approaches into routine nephrogenetic practice. Full article

Background: Healthy dietary behaviours are essential for maintaining health, functional independence, and quality of life in later life. Family members are a key source of social support for community-dwelling older adults, yet the ways in which family support shapes older adults’ dietary behaviours, particularly among those who retain autonomy, remain insufficiently synthesized. Therefore, this review aims to map how family support influences dietary behaviours among community-dwelling older adults by examining the forms, roles, and contextual influences of family support within a Social Support Theory framework. Methods: Following Joanna Briggs Institute guidance and PRISMA-ScR reporting standards, we conducted a scoping review of empirical studies published in English or Chinese. Searches were conducted across PubMed, CINAHL, PsycINFO, Web of Science, and Scopus from inception to 2025. Quantitative and qualitative evidence was synthesised using a convergent–segregated mixed-methods approach. Qualitative findings were deductively mapped to instrumental, informational, emotional, and esteem support domains. Results: Nineteen studies were included. Quantitative evidence indicated that family support, particularly shared meal preparation, joint dietary adherence, and autonomy-supportive encouragement, was generally associated with better diet quality, dietary adherence, and nutritional outcomes. Qualitative findings showed that the influence of family support depended on relationship dynamics and contextual factors, including communication patterns, autonomy negotiation, shared responsibility, and cultural expectations. Conclusions: Family support plays a multifaceted and context-dependent role in shaping dietary behaviours among community-dwelling older adults. These findings can inform the development of family-inclusive strategies and interventions that promote healthy dietary behaviours while respecting older adults’ autonomy and relational contexts. Full article

Feedback is an integral part of the learning process. However, delivering feedback effectively remains challenging, particularly within massified higher education systems that are characterised by large cohorts and increasingly diverse student populations. The emergence of generative artificial intelligence (GenAI) enables new ways of embedding feedback into educational offerings, some of which may be highly beneficial. In this paper, we introduce Abundant Feedback as a conceptual lens for examining the new capabilities that may be enabled by GenAI. We present a four-dimensional framework identifying the dimensions of GenAI feedback as abundance of Volume, of Availability, of Relevance and of Character. Through a systematic literature search, we describe how these dimensions manifest in recent empirical studies, and identify two educational domains, Computer Programming and Foreign Languages, as early adopters of AI-generated feedback. Beyond merely digitising existing scarce feedback processes, we discuss the emergence of new learner-driven feedback practices that are enabled by abundance, that both stimulate and demand student feedback literacy. Our multi-dimension abundance framework provides a lens, as well as the vocabulary and conceptual tools, to guide the implementation of GenAI feedback in ways that help realise the potential of artificial intelligence to enhance student learning. Full article

Background: Mitochondria are multifunctional organelles that play a central role in maintaining cellular homeostasis by regulating energy metabolism, reactive oxygen species (ROS) generation, ion homeostasis, and apoptotic signaling. Dynamic processes such as mitochondrial fission, fusion, and intracellular trafficking enable cells to adapt to metabolic and environmental stress. Growing evidence indicates that dysregulation of these processes is a hallmark of cancer, contributing to metabolic reprogramming, redox imbalance, evasion of apoptosis, and disease progression. This narrative review aims to discuss the role of mitochondrial alterations in the pathophysiology of chronic myeloid leukemia (CML) and their potential therapeutic implications. Methods: Original research articles published between 2010 and 2025 were considered in this narrative review. The selected studies were critically discussed and categorized into three principal thematic domains: mitochondrial regulation of redox homeostasis, metabolic rewiring, and control of cell death pathways. Evidence was synthesized to elucidate the contribution of mitochondrial dysfunction to CML initiation, progression, and therapeutic resistance. Results: The reviewed studies highlight how mitochondrial abnormalities play a pivotal role in BCR-ABL1-driven leukemogenesis. Alterations in mitochondrial metabolism and ROS signaling support sustained proliferative signaling, promote genomic instability, and facilitate resistance to apoptosis. In addition, mitochondrial adaptations contribute to resistance to tyrosine kinase inhibitors (TKIs) and are essential for the persistence and survival of leukemic stem cells. Conclusions: Mitochondria emerge as central regulators of CML pathobiology. Therapeutic strategies targeting mitochondrial metabolism, redox homeostasis, and apoptotic signaling pathways represent promising approaches to overcoming TKI resistance and may improve clinical outcomes for patients with CML. Full article

Due to the complex mesostructure of calcareous sand, accurately predicting the mechanical response of Asphalt-Bonded Calcareous Sand (ABCS) is extremely challenging. This study pioneers the development of a mesoscale model for ABCS that explicitly incorporates the Interfacial Transition Zone (ITZ) via a random particle algorithm. To overcome the efficiency bottlenecks of traditional time-domain integration, this study establishes a mesoscale framework coupling a random polygonal aggregate algorithm with direct Steady-State Dynamics (SSD) analysis. A major advantage of this framework is its capacity for large-scale parametric sensitivity analysis; herein, 920 independent mesoscale models were generated and rapidly solved across the broadband frequency domain. The framework was rigorously validated, demonstrating high predictive accuracy for both the baseline calibration and an independent 12% asphalt content mixture (baseline R² = 0.99, MAPE = 6.94%; independent validation R² = 0.96, MAPE = 9.73%). Notably, the SSD approach completes calculations (10⁻³ to 10³ Hz) for 10 massive 300 mm RVEs in just 6.5 min. Leveraging this high-throughput capability, the extensive parametric analysis reveals that variations in maximum aggregate size negligibly impact the dynamic modulus under a constant volume fraction. Conversely, an optimal Interfacial Transition Zone (ITZ) thickness of ~75 µm was identified, representing a physical equilibrium between interfacial reinforcement and bulk binder cohesion. Furthermore, an analytical RVE size criterion of 1.7–5.3 times the maximum aggregate size is proposed to satisfy a 5% engineering error tolerance, providing a highly efficient numerical tool for the virtual mix design of reef pavements. Full article