Search Results (24,838)

Geographical authentication of medicinal plant materials remains challenging because multispectral variables are often highly collinear and sample grouping can complicate reliable model validation. Existing correlation-based feature-selection strategies also require careful adaptation to multiclass problems to avoid artificial ordering of class labels and information leakage during model development. Therefore, this study aimed to develop a compact and leakage-controlled multispectral learning framework for geographical-origin discrimination. This study analyzed 800 physical Pinellia ternata samples from Gansu Xihe, Sichuan Neijiang, Sichuan Chengdu, and Chongqing Dianjiang (200 samples per origin). Each physical sample was represented by 31 mean grayscale intensities calculated from Otsu-segmented multispectral regions of interest. A Pearson-correlation-guided deep multilayer perceptron (PCG-DeepMLP) was constructed by estimating one-vs-rest band relevance and inter-band redundancy only within the training data. The key methodological innovation is a unified multiclass-aware, relevance–redundancy spectral-learning framework in which class-specific one-vs-rest Pearson relevance is coupled with inter-band redundancy control and embedded within leakage-controlled grouped model development. By learning the spectral subset exclusively from each training partition before nonlinear classification, the framework produces compact and complementary multispectral signatures while preserving multiclass structure and strict independence of held-out groups. Model and feature-selection settings were chosen by three-fold grouped cross-validation within each training partition. PCG-DeepMLP retained 9–21 bands and achieved the highest mean accuracy (0.9812 ± 0.0135), macro-F1 (0.9812 ± 0.0135), Matthews correlation coefficient (MCC; 0.9752 ± 0.0179), and macro-AUC (0.9994 ± 0.0006) among seven models. Its macro-F1 was higher than that of 1D-CNN, 1D-ResNet, full-band MLP, PLS-DA, and random forest after Holm correction. Performance was estimated through a strict nested group-wise internal validation scheme, with every outer test fold remaining isolated from feature selection, preprocessing, and model optimization. These findings demonstrate that multiclass-aware relevance–redundancy learning can retain complementary Pinellia ternata origin-discriminative information in a compact and stable spectral representation, enabling accurate geographical-origin authentication while providing a principled basis for reduced-channel acquisition and future independent multi-batch validation. Full article

During long-term monitoring, data missing often happens due to environmental interference, sensor malfunction or transmission problems, which will threaten the effectiveness of structural health monitoring. This study proposes an anchor-constrained residual stacking method, which can improve data reconstruction performance relative to individual models across diverse missing patterns while reducing the risk of performance degradation associated with unconstrained ensemble learning. The method adopts a two-level stacking framework in which heterogeneous base learners generate candidate reconstructed data and grouped out-of-fold predictions are used to select an anchor learner for different missing patterns. A residual meta-learner then learns complementary residual information relative to the anchor learner, while validation-gated fusion regulates residual correction and final fusion based on reserved validation data, reducing unreliable fusion contributions. The effectiveness of the proposed method is verified using monitoring data with different missing patterns from a steel stringer bridge under multiple structural states. It achieves a coefficient of determination ($R^2$) of 0.9020, together with the lowest relative root mean square error (RRMSE) and mean absolute error (MAE) among the compared methods. Operational modal analysis further confirms that the method preserves the main dynamic characteristics of structural responses, supporting reliable reconstruction across diverse missing patterns and multiple structural states. Full article

Accurate weld keypoint localization is an important visual perception task for robotic welding, weld tracking, and intelligent manufacturing. However, weld keypoints in line-structured light images are usually small, weakly textured, and easily affected by reflections, noise, and spurious laser stripes. These factors make it difficult for lightweight detection models to balance localization accuracy and computational efficiency. To address this problem, this paper proposes Weld-FHG-YOLO, a lightweight multi-frequency feature fusion network for weld keypoint localization. The proposed model is built on the You Only Look Once version 11 nano (YOLO11n) framework and focuses on optimizing feature fusion and scale transformation in the Neck. Specifically, FasterC3K2 is introduced to replace the original C3K2 modules in the Neck, thereby reducing redundant computation during multi-scale feature fusion. In addition, a Haar Wavelet Decomposition and Group Shuffle Convolution (HWD-GSConv) downsampling fusion module is designed, in which Haar wavelet decomposition preserves low-frequency structural information and high-frequency details, while GSConv performs lightweight multi-frequency feature fusion. Experimental results show that Weld-FHG-YOLO achieves 2.301 M parameters and 6.016 GFLOPs, which are 11.26% and 6.83% lower than those of YOLO11n, respectively. Meanwhile, mAP@0.5:0.95 increases from 0.7337 to 0.7901, the Mean Center Error (MCE) decreases from 2.254 px to 2.131 px, and the CPU inference speed increases from 13.69 to 14.77 frames per second (FPS). These results indicate that the proposed method improves strict localization accuracy and localization stability while maintaining a lightweight computational profile, providing a practical visual perception approach for weld keypoint localization in resource-constrained intelligent manufacturing scenarios. Full article

A pipe-in-pipe structure with a buoyancy damping layer is designed to reduce pipeline weight and suppress vortex-induced vibration. This structure avoids structural pre-stress problems caused by external buoys and mitigates severe local vibration. The four-degree-of-freedom motion equation of the novel pipe-in-pipe system is established, and a corresponding numerical program is compiled. Bidirectional fluid–structure interaction (FSI) simulations are performed to verify the vibration reduction performance of the proposed structure. The auxiliary outer pipe alters surrounding flow field characteristics, generates distinct hydrodynamic forces and modifies structural vibration behaviors. Massive simulation data demonstrate that the buoyancy damping layer of the pipe-in-pipe structure can adapt to hydrodynamic forces with various characteristics and achieves favorable broadband vibration suppression and energy absorption. For all layout parameters, the inner damping layer exhibits an outstanding suppression effect on cross-flow vibration of the cylinder-auxiliary pipe system. The inline vibration amplitude of the pipe-in-pipe-auxiliary pipe system is consistently less than 0.08D and can be neglected. Its vibration frequency is close to that of the cross-flow direction and only half that of the single-layer pipe. The inner damping layer also affects the fluid–structure interaction between the pipeline system and external flow field, resulting in variations of the vorticity field. Full article

Agricultural ammonium pollution from farmland drainage and low-value crop straw utilization are two critical rural environmental problems that cannot be solved by single remediation approaches. Herein, a novel composite was prepared by immobilizing effective microorganisms (EM) on corn straw biochar to construct a synergistic adsorption–biodegradation system, and its nitrogen removal mechanism was systematically investigated at structural and molecular levels. Metagenomic analysis detected a complete set of heterotrophic nitrification–aerobic denitrification (HN-AD) functional genes (amoA, hao, napA, nirK, norB, nosZ) in the isolated strain Bacillus thuringiensis A1, revealing the genetic potential of this strain for ammonium biodegradation. EM modification optimized biochar pore structure and increased the equilibrium adsorption capacity to 1.215 mg/g, which was 66.4% higher than that of pristine biochar (0.73 mg/g). Sterilization control tests indicated that physicochemical adsorption occupied the dominant position in ammonium removal, while microbial biodegradation acted as an auxiliary removal pathway. Importantly, the synergistic relationship between the two pathways should be interpreted cautiously, since autoclaving may subtly alter biochar physicochemical properties, and direct paired characterization of viable composites before and after sterilization is technically unavailable. Kinetic and thermodynamic results further validated the improved adsorption performance after modification. Overall, EM immobilization promoted ammonium adsorption via pore optimization, while pore-confined microbes achieved sustainable HN-AD biotransformation, jointly realizing synergistic nitrogen removal. This study provides a mechanistic reference for the optimized design and application of biochar–microbe composites in agricultural nitrogen pollution control. Full article

Federated learning enables model training in edge and distributed service environments without directly sharing raw data. In long-running edge federated services, fixed collaboration structures may become inefficient under workload drift, whereas frequent regrouping can incur migration overhead, group churn, and additional privacy-budget consumption. This paper studies budget-prioritized dynamic regrouping under workload drift, privacy-budget constraints, and migration or reconfiguration cost. We formulate a dynamic regrouping problem that jointly captures workload pressure, remaining privacy budget, service utility, and regrouping cost. We propose Budget-Prioritized Dynamic Regrouping (BP-DR), a triggered local method that evaluates single-node candidate operations and commits at most one regrouping operation per time slot. A candidate is accepted only when it is privacy-budget feasible and its utility improvement exceeds a threshold combining an anti-oscillation margin, migration or reconfiguration cost, and privacy-budget opportunity cost. We derive this trigger from a one-step local comparison and establish its monotonicity with respect to migration or reconfiguration cost and remaining privacy budget. Trace-driven experiments based on Alibaba Cluster Trace 2018 show that BP-DR maintains competitive migration-adjusted utility while controlling regrouping activity across dynamic and stress-test settings. FLamby Fed-Heart-Disease validation further shows similar learning performance across the compared methods while demonstrating the integration of BP-DR with group-aware federated training. Full article

Embedding an evolutionary algorithm (EA) into control structures offers an effective solution for specific control problems. Often, it predicts the best control values using a process model (PM). The primary limitation is its high computational time. Our work addresses optimal control problems (OCPs) with a final cost, using receding horizon control (RHC) with an EA as a predictor. This work is a continuation of a previous article, in which the EA predictor was replaced with a multilinear regression-based predictor. Our objective is to propose a predictor based on regression neural networks (RNNs) that emulates the behavior of the (EA, PM) couple. A number of closed-loop simulations using the existing EA controller produce sequences of optimal control values and corresponding state values, which are stored in a data structure. Datasets for each sampling period are derived from these data and are used to train RNN objects employing a unique RNN model. The model, which is an “optimizable” RNN plus the list of hyperparameters preset before optimization, is determined after a thorough analysis of possible candidates using a MATLAB R2025b application. The presented method of constructing an RNN predictor is the main contribution. Algorithms for (a) constructing the sequence of RNN objects and (b) simulating the closed loop are also proposed. A case study illustrates our method. The RNN predictor successfully emulated the (EA, PM) couple: (a) the control-loop dynamics were nearly identical; (b) the performance indices were essentially the same; and (c) the execution time of the controller significantly decreased from 38 to 0.054 s, demonstrating that RHC can be applied more broadly. Full article

Class imbalance is prevalent in real-world datasets. Minority samples are far fewer than majority samples. Traditional classifier design typically assumes balanced data, which causes classifiers to favor the majority class when faced with imbalanced datasets. Thus, there are high misclassification costs for minority classes in critical fields like healthcare and finance. Most existing oversampling methods for handling imbalance problems, such as SMOTE (the synthetic minority oversampling technique), suffer from limitations like noise sensitivity, failure to consider minority-class sub-cluster structures, and poor adaptability to the heterogeneity of sample distributions. This article addresses these issues by proposing a novel oversampling algorithm: the adaptive weighting–synthetic minority oversampling technique (AW-SMOTE). It consists of three progressive stages. In the first stage, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering identifies the distribution structure of the minority samples. It identifies potential sub-clusters while removing noise interference. A clear data foundation for subsequent sampling is established. In the second stage, the most representative boundary sample in each cluster is used to evaluate the weight of each cluster. The total number of synthetic samples is allocated to different clusters according to their weights. This provides global sample enhancement support. In the third stage, adaptive sample generation is performed within each cluster. It combines the two perspectives of boundary tightness and local density. The sigmoid function is used to dynamically adjust the weight ratio. Finally, new samples are synthesized in key regions to both preserve distribution characteristics and enhance discriminability in classification. Through experiments on standard datasets from the KEEL repository, the feasibility and effectiveness of this algorithm are demonstrated. Full article

Background/Objectives: Pediatric lower urinary tract symptoms and bladder–bowel dysfunction are common and clinically related problems recognized within the ICCS framework, but no bibliometric study has previously mapped this broad field comprehensively. This study aimed to map the global structure, thematic evolution, and emerging research trends in pediatric lower urinary tract dysfunction using bibliometric methods. Methods: A total of 3004 publications published between 1945 and May 2026 and retrieved from the Web of Science Core Collection were analyzed. Keyword co-occurrence networks, country co-authorship networks, and document citation networks were constructed using VOSviewer (version 1.6.20)). Results: Annual publication counts first exceeded 100 in 2013 and remained largely at or above this level thereafter, peaking at 135 in 2020, with a subsequent plateau. The USA (26.1%), Turkiye (10.5%), and the United Kingdom (6.5%) were the most productive countries. Harvard University led institutional output (n = 131). Keyword network analysis identified 15 clusters, the most coherent of which centered on enuresis pathophysiology/pharmacology, overactive bladder treatment, sleep medicine, and core urinary and bowel symptoms. Overlay visualization identified artificial intelligence (~2024), telemedicine (~2022), and autism spectrum disorder (~2021) as the most recent emerging research themes. Conclusions: This is the first bibliometric analysis to map global pediatric lower urinary tract dysfunction research across the broader field. Artificial intelligence, telemedicine, and autism spectrum disorder emerged as the most recent research themes, while substantial geographic gaps remained in sub-Saharan Africa and South Asia. Full article

Double negative T cells (DNT cells) are a rare T cell population involved in immune regulation, inflammation, autoimmunity, transplantation, and tumor immunity. However, their low abundance and dispersed distribution across tissues have limited a systematic understanding of their cellular organization and transcriptional diversity. To address this problem, we developed an integrative single cell framework to reconstruct the mouse DNT cell landscape across tissues using multiple public single cell RNA sequencing datasets. Using transcriptomic criteria, we identified and integrated 7984 mouse RNA defined DNT cells. The integrated population was resolved into multiple transcriptionally distinct clusters. These clusters were organized into naive like, proinflammatory, cytotoxic, proliferative, and myeloid associated states, revealing that mouse RNA defined DNT cells constitute a highly heterogeneous yet structured transcriptional compartment. Integrated downstream analyses indicated transcriptional relationships among these subsets. Naive like populations showed inferred transcriptional relationships with inflammatory and cytotoxic programs, while virtual knockout and intercellular communication analyses identified distinct predicted regulatory and signaling features across states. To place these mouse DNT associated genes in a human disease context, selected genes were mapped to their corresponding human homologs and examined using TCGA pan cancer transcriptomic data. Together, these findings define a cross tissue transcriptional framework for mouse RNA defined DNT cells and provide a basis for further evaluating the conservation and relevance of these transcriptional programs in human DNT biology. More broadly, this study provides a generalizable integrative strategy for reconstructing and characterizing rare immune cell populations that are insufficiently represented in individual single cell datasets. Full article

Reliable electricity supply is essential for healthcare institutions, particularly where grid instability can disrupt service continuity, compromise patient safety, and affect the operation of critical medical equipment. In this context, selecting an appropriate photovoltaic (PV) proposal is a complex decision problem that requires assessing technical, economic, environmental, and regulatory factors jointly. This study develops a hybrid multi-criteria decision-making framework that integrates the Fuzzy Analytic Hierarchy Process (FAHP), the Decision-Making Trial and Evaluation Laboratory (DEMATEL), and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to support PV proposal selection in healthcare institutions. The framework was applied to four competing proposals for a hospital case study in Barranquilla, Colombia. After integrating FAHP and DEMATEL, the economic, technical, and environmental criteria received balanced interdependence-adjusted weights of 0.324, 0.337, and 0.338, respectively. At the same time, DEMATEL identified the technical dimension as the main net influencing dimension within the expert-elicited influence network. The final ranking placed Proposal 1 first, followed by Proposal 4, Proposal 2, and Proposal 3, with closeness coefficients of 0.530, 0.518, 0.498, and 0.492, respectively. Additional comparative analysis showed that omitting DEMATEL changed the winning alternative, whereas preserving the FAHP–DEMATEL weighting structure and replacing TOPSIS with MARCOS yielded the same ranking. Robustness analyses further showed that the ranking remained stable in most supplier-exclusion and leave-one-expert-out scenarios. In contrast, bootstrap-based probabilistic sensitivity analysis showed that Proposal 1 ranked first in 96.2% of the replications. These results support the practical usefulness of the proposed framework for decision-making in healthcare energy planning. Full article

Civic education today is shaped by access to information via social media, where disinformation and polarisation pose significant challenges. In this context, initial teacher training must foster critical thinking and democratic competence to prepare future teachers who are committed to addressing the issues in their local communities. This study analyses which local eco-social issues are identified by students on the Bachelor’s Degree in Primary Education at the University of Extremadura, how these are represented on the social media platforms they use, and what teaching strategies they propose to address them in schools. Using an action-research approach, an intervention involving 164 students was developed, structured in three phases: identification of local issues, analysis of their presence on social media, and the design of teaching proposals using educational resource packs tailored to primary education. The results reveal differences between direct observation of the problems and their treatment on social media. Furthermore, the proposal demonstrates that working with relevant social issues, critically analysing social media and designing teaching resource packs promote reflection, critical awareness and the ability to apply teaching strategies among future teachers. Full article

Background/Objectives: Structured cytopathology reporting systems attach an evidence-based risk of malignancy (ROM) to every diagnostic category for the thyroid, salivary gland, urinary tract, lung and other sites. None exists for the oral cavity, and the ROM of an oral brush cytology category has not, to our knowledge, been reported. Methods: All oral liquid-based cytology specimens accessioned at one tertiary dental hospital between 2015 and 2024 were reviewed; those with a site-concordant histopathological diagnosis obtained within 90 days formed the analytic cohort (n = 137). Cytological reports were recovered from primary laboratory records and assigned to a five-tier ordinal scale. The target condition was malignancy or high-grade dysplasia, and diagnostic indices were calculated at every ordinal threshold with Wilson 95% confidence intervals. Results: ROM rose monotonically: 15.6% (95% CI 8.7–26.4) for negative, 53.3% (30.1–75.2) for atypical favoring reactive, 76.9% (61.7–87.4) for atypical cells of undetermined significance (ACUS), 87.5% (52.9–97.8) for atypical favoring neoplastic and 100% (74.1–100) for malignancy. The ACUS value far exceeded that of the similarly named tier in the thyroid (22%) and salivary gland (30.5%) systems and fell within the range spanned by their suspicious-for-malignancy tiers (74% and 83.8%). Malignancy was never reported in a histologically benign or low-grade lesions. Sensitivity at the ACUS threshold was 25.0% in keratotic against 79.3% in non-keratotic lesions (p = 0.004). Conclusions: These are, to our knowledge, the first category-specific ROM estimates for oral brush cytology, and they identify a nomenclature problem: a category named for the pathologist’s uncertainty carried a risk equivalent to suspicious for malignancy elsewhere; while a negative report was followed by disease in one lesion in six. Any oral reporting system should attach an explicit ROM to each category; our exploratory data further suggest that the surface keratinization of the sampled lesion should be recorded. Full article

Sustainable urban freight logistics requires routing decisions that jointly account for operating cost, vehicle technology, low-emission-zone (LEZ) access, terrain-sensitive energy use, and battery feasibility. This study develops a policy–terrain-coupled mixed-fleet routing framework integrating LEZ exposure, system-level carbon settlement, terrain-sensitive energy consumption, electric-vehicle (EV) battery feasibility, and route-level EV/internal-combustion-engine vehicle reassignment within a unified daily total operational cost (DTOC) evaluator. An adaptive large-neighborhood search (ALNS) procedure reconstructs feasible routes, while vehicle type is re-evaluated through counterfactual comparison of the complete system objective. The main experiments use 60 enhanced Gehring–Homberger benchmark-derived scenarios and 20 independent seeds, supplemented by ablation, carbon-price, EV-fixed-cost, heuristic-weight, convergence, and customer-scale scalability analyses. The ALNS-based framework achieves the lowest mean DTOC among the tested procedures, albeit with higher runtime. Policy and terrain information alter modeled fleet composition, with topology-dependent cost effects. Lower EV fixed costs consistently increase EV share, whereas carbon-price effects vary across network structures. All runs in the additional 200–1000-customer tests were feasible, although runtime increased with problem size. London- and Madrid-informed cases are treated as archetypes rather than as real-world validation cases. These results provide a basis for scenario screening and comparative planning of policy–terrain interactions before city-specific calibration and deployment. Full article

Prostate cancer is the prototypical immunologically cold solid tumor, with objective response rates of only 3–5% to immune checkpoint inhibitors in unselected metastatic castration-resistant disease and an estimated 89.8% of tumors classified as immunologically ignorant. Three convergent features sustain this phenotype: low tumor mutational burden with defective Major Histocompatibility Complex class I antigen presentation; an immunosuppressive microenvironment dominated by regulatory T-cells, myeloid-derived suppressor cells, and M2 macrophages; and dense stromal and vascular barriers that exclude effector lymphocytes. Across these mechanisms, one compartment has received disproportionately little attention: the lymphatic system. Tumor-associated lymphatic remodeling driven by vascular endothelial growth factor C and D produces structurally abnormal vessels that impair antigen and dendritic cell trafficking to tumor-draining lymph nodes; the lymph nodes themselves are reprogrammed to a tolerogenic state by lymphatic endothelial cells expressing programmed death-ligand 1 and lacking costimulation and by regulatory T-cells that suppress effector egress. This review synthesizes evidence that in prostate cancer, immune coldness reflects not merely a problem of checkpoint engagement with the tumor but a failure of antigen trafficking at the lymphatic interface. We discuss therapeutic strategies that target this axis, such as lymphangiogenesis-inducing vaccines, lymph-node-directed checkpoint delivery, induction of intratumoral tertiary lymphoid structures and stromal reprogramming, as complements to existing immunotherapy. Targeting trafficking, not only checkpoints, may be the prerequisite for converting PCa from cold to hot. Full article