Browse Articles

A legal judgment prediction model based on knowledge fusion and dependency masking

Legal Judgment Prediction (LJP) is a core task in Legal AI systems, which aims to predict law articles, charges, and term-of-penalty from case facts. While existing deep-learning-based LJP approaches for civil law systems have achieved certain progress, they still suffer from two key limitations: (1) insufficient deep understanding and effective utilization of external judicial knowledge; and (2) the lack of effective strategies to filter out erroneous dependency information in multi-task LJP frameworks. To address these challenges, we propose a legal judgment prediction model based on knowledge fusion and dependency masking. Specifically, we first integrate a CNN-based local semantic refinement component into the existing BERT-based legal knowledge extraction method, thereby enabling the model to further extract the core knowledge embedded in judicial documents. Then, we introduce differential attention to reduce noise in conventional attention fusion methods and help the model locate key information in case facts more accurately. Furthermore, we propose a multi-task dependency information masking mechanism to accurately identify and filter erroneous dependency information for multi-task LJP methods. Experiments conducted on real-world datasets demonstrate the superiority of our proposed model. This code is available online at https://github.com/PaperCode-GNU/KFTM .

Transdiagnostic dimensions of psychopathology in chronic tinnitus patients with and without hearing loss

Abstract Chronic tinnitus is frequently described as occurring with psychiatric comorbidities. However, this diagnosis-based framing may obscure underlying dimensions of psychological vulnerability in individuals both with and without measurable hearing loss (HL). This study thus aimed to identify transdiagnostic dimensions of psychopathology in patients with chronic tinnitus and examine their relevance for tinnitus-related distress (TRD) in patients with or without HL. In a clinical sample of N = 678 chronic tinnitus patients, pure-tone audiometry determined HL status and self-report screeners assessed psychiatric symptom profiles. Multiple regression models tested diagnostic predictors of TRD and diagnosis × HL interactions. Tetrachoric correlations among binary ‘diagnosis pointers’ were subjected to principal axis factor analysis to derive latent dimensions, which were then conceptually linked to the Hierarchical Taxonomy of Psychopathology (HiTOP). Most patients (96%) endorsed at least one diagnosis pointer (median = 5); over half (52.1%) had no identifiable HL. Patients without HL exhibited higher rates of anxiety and substance-related symptoms. TRD was significantly predicted by diagnosis pointers for major depression, agoraphobia, health anxiety, anorexia nervosa, and psychosis. Factor analysis identified three dimensions: Internalizing psychopathology, harmful substance use, and fear-related (social) perceptions. Internalizing psychopathology predicted TRD. HL status was associated with harmful substance use but not with psychological predictors of distress. Findings suggest that chronic tinnitus indexes broad psychological vulnerability rather than discrete psychiatric comorbidities. Whilst inner ear damage may trigger tinnitus onset in patients with HL, persistence across patients—particularly those without HL—suggests a psychological substrate underlying symptom chronicity, distress, and diagnostic covariation. Clinically, the findings support a dimensional, formulation-based approach that situates tinnitus within patients’ broader emotional functioning and foregrounds transdiagnostic processes over discrete diagnoses.

Deep learning with satellite images enables high-resolution income estimation: A case study of Buenos Aires

High-resolution income data is crucial for informing policy decisions as it allows policymakers to better understand the distribution of wealth and poverty. However, obtaining this information is often cost-prohibitive, especially in developing countries. We evaluate the potential of using high-resolution satellite imagery and machine learning techniques to create income maps with a high level of geographic detail. We train a neural network with satellite images from the Metropolitan Area of Buenos Aires (Argentina) and 2010 census data to estimate per capita income at a 50x50 meter resolution for 2013, 2018 and 2022. The model, based on the EfficientNetV2 architecture, demonstrates strong predictive accuracy for household incomes ( R 2  = 0.878), achieving a spatial resolution over 20 times finer than existing methods in the literature. The model also allows estimating income maps for arbitrary images, and can therefore be applied at any point in time. Our approach opens up new possibilities for generating highly detailed data, which can be used to assess public policies at a local level, target social programs more effectively, and address information gaps in areas where traditional data collection methods are lacking.

Strengthening disaster preparedness and health security in Niger state, Nigeria through a WHO STAR–based multi-hazard risk assessment

Abstract Niger State in central Nigeria faces a range of natural, biological, and security hazards. To inform preparedness and health security planning, a multi-hazard risk assessment was conducted using WHO’s Strategic Tool for Assessing Risks (STAR), this is one of the first applications of WHO STAR at a state level in Nigeria. A cross-sectional study was conducted using the WHO STAR. Stakeholders involved identified hazards across natural, biological, technological, and societal domains through review of surveillance, disaster, and meteorological data. Hazards were scored for likelihood, impact, vulnerability, and coping capacity, with composite risk indices used to rank and categorize them. Priority hazards were further analysed for seasonality and geographic distribution, and findings validated through consensus. Eighteen major hazards were identified, spanning biological, environmental, and societal. Seven hazards emerged as very high risk, notably flooding, banditry/kidnapping. Six were high risk (e.g. fire outbreaks), four moderate (e.g. acute flaccid paralysis), and one low risk (diphtheria). Six hazards showed clear seasonal patterns. Priority hazards were further examined for geographic distribution and validated through consensus. The STAR assessment produced an evidence-based risk profile highlighting flooding, banditry/kidnapping, boat mishaps, cholera, and rain/windstorms as the most critical hazards. Actionable recommendations were developed to support preparedness, mitigation, and response efforts across sectors. The findings offer a structured basis for strengthening disaster risk governance and can inform the development and implementation of Niger state’s emergency preparedness plans.

Testing the utility of the first step of system evaluation theory in creating a system map of care for cardiac amyloidosis early detection: A case study

Background Heart failure is a clinical syndrome resulting from numerous pathological conditions. One cause of heart failure, transthyretin cardiac amyloidosis, presents insidiously with common and seemingly unrelated symptoms. New treatments for cardiac amyloidosis are available that extend and improve life. However, providers are not testing patients for transthyretin cardiac amyloidosis. We took a systems science approach to explore the system of care for transthyretin cardiac amyloidosis testing by depicting the healthcare system from patient presentation to treatment. Our goal was to define an ideal healthcare system to improve the uptake of testing protocols and enhance patient outcomes. Methods We assembled clinicians, researchers, and patients to participate in a co-design workshop using the first step of System Evaluation Theory to define an ideal testing and diagnostic protocol using transthyretin cardiac amyloidosis as a case study. We tasked workshop attendees with defining the patient and clinician journey from symptom presentation to diagnosis. We generated a system map using a qualitative matrix analysis of a transcript of the workshop discussion. Results The matrix analysis organized input from all stakeholders, allowing for the creation of a system map that reveals the complexity of the transthyretin cardiac amyloidosis testing process and potential implementation strategies to improve the efficiency and effectiveness of the system. This methodology successfully yielded generalizable elements of a testing protocol and testable strategies to facilitate the implementation of a protocol adapted to fit local site needs. Conclusions The substeps outlined within System Evaluation Theory Step 1 helped identify an ideal system for testing and diagnosing transthyretin cardiac amyloidosis care that could be applied to specific settings to identify, improve, and implement protocols for other complex diseases.

Optimization study on transverse mining zoning during the capacity expansion stage of nearly horizontal open-pit coal mines

Alzheimer’s disease prediction via an explainable CNN using genetic algorithm and SHAP values

Convolutional neural networks (CNNs) are widely recognized for their high precision in image classification. Nevertheless, the lack of transparency in these black-box models raises concerns in sensitive domains such as healthcare, where understanding the knowledge acquired to derive outcomes can be challenging. To address this concern, several strategies within the field of explainable AI (XAI) have been developed to enhance model interpretability. This study introduces a novel XAI technique, GASHAP, which integrates a genetic algorithm (GA) with SHapley Additive exPlanations (SHAP) to improve the explainability of our 3D convolutional neural network (3D-CNN) model. The model is designed to classify magnetic resonance imaging (MRI) brain scans of individuals with Alzheimer’s disease and cognitively normal controls. Deep SHAP, a widely used XAI technique, facilitates the understanding of the influence exerted by various voxels on the final classification outcome (Lundberg SM, Lee SI. A unified approach to interpreting model predictions. In: Advances in Neural Information Processing Systems, 2017. 4765–74. https://doi.org/10.5555/3295222.3295230 ). However, voxel-level representation alone lacks interpretive clarity. Therefore, the objective of this study is to provide findings at the level of anatomically defined brain regions. Critical regions are identified by leveraging their SHAP values, followed by the application of a genetic algorithm to generate a definitive mask highlighting the most significant regions for Alzheimer’s disease diagnosis (Shahamat H, Saniee Abadeh M. Brain MRI analysis using a deep learning based evolutionary approach. Neural Netw. 2020;126:218–34. https://doi.org/10.1016/j.neunet.2020.03.017 PMID: 32259762). The research commenced by implementing a 3D-CNN for MRI image classification. Subsequently, the GASHAP technique was applied to enhance model transparency. The final result is a brain mask that delineates the pertinent regions crucial for Alzheimer’s disease diagnosis. Finally, a comparative analysis is conducted between our findings and those of previous studies.

Classifying retinal images via vascular-optic disc cross-segmentation and attentive feature selection

Nest attributes influence choice accuracy, but not decision latency in acorn ants

Decision making can have significant fitness consequences across various aspects of animal life. For acorn ants, Temnothorax curvispinosus , choosing a new nest quickly and accurately can affect the survival and fitness of the whole colony. When emigrating, ants consider several nest attributes such as cavity shape, height, and brightness. Ants may benefit from having more attributes differentiating potential nests only if they can quickly and accurately assess all possible attributes and make well-informed decisions. Here, we asked if the number and type of attributes differentiating potential nests affected the accuracy and latency of colony decision-making. We used pair-wise tests, where potential nests differed in one to three attributes, with one nest within the pair considered less optimal. We recorded which nest colonies chose and the time it took to make the decision. We found that accuracy increased with the number of attributes, particularly when nest brightness was manipulated, indicating that increasing the number of attributes may help facilitate nest-site selection. We also found that the degree of difference did not affect the decision-making latency, suggesting that ant colonies searching for a new nest might be constrained temporally when selecting a new nest site.

Automated 4D flow MRI pipeline for the quantification of advanced hemodynamic parameters in the left atrium

Abstract The left atrium (LA) plays a pivotal role in modulating left ventricular filling, yet its hemodynamics remain poorly understood due to the limitations of conventional ultrasound analysis. Four-dimensional flow magnetic resonance imaging (4D Flow MRI) holds promise for enhancing our understanding of atrial hemodynamics, but its analysis is hindered by the inherently low velocities within the chamber and the modest spatial resolution of 4D Flow MRI. Heterogeneity in acquisition protocols and MRI vendors, and the lack of standardized computational frameworks further complicates the creation of large, comparable datasets needed to assess the prognostic value of hemodynamic markers provided by 4D Flow MRI. To address these challenges, we introduce a computational framework tailored to the analysis of 4D Flow MRI in the LA, enabling the qualitative and quantitative analysis of advanced hemodynamic parameters (e.g., kinetic energy, vorticity, and pressure). We applied this framework to a diverse cohort spanning different degrees of left ventricular diastolic dysfunction to investigate the prognostic potential of these metrics. Our framework proved robustness across multicenter data of varying quality, producing high-accuracy automated segmentations. Notably, our findings show that 4D Flow MRI-derived parameters provide superior differentiation between healthy and pathological states than those available to conventional hemodynamic analysis tools.

Rhodopsin‐Mimicking Reversible Photo‐Switchable Chloride Channels Based on Azobenzene‐Appended <i>Semiaza</i> ‐Bambusurils for Light‐Controlled Ion Transport and Cancer Cell Apoptosis

Abstract The ability to control ion transport across membranes in living systems by stimulus‐responsive natural channels, such as channelrhodopsins and their mimics, is a revolutionary tool for understanding biological processes. Herein, we demonstrate a new class of azo‐functionalized bambusurils (azo‐BUs) that act as efficient, photo‐switchable anion channels capable of modulating chloride flux across lipid membranes and within cellular environments. The ( E )‐isomer exhibits pronounced chloride transport activity, which can be reversibly toggled via light‐induced isomerization, enabling precise spatiotemporal control. Mechanistic studies reveal that the ( E )‐form induces apoptosis through mitochondrial membrane depolarization, reactive oxygen spieces (ROS) generation, and cytochrome c release, while also disrupting lysosomal acidification via H⁺/Cl − cotransport. This dual perturbation of cytosolic and lysosomal ion homeostasis underscores the compound's multifaceted cytotoxic mechanism. In contrast, the ( Z )‐isomer displayed minimal transport activity and negligible cytotoxicity, reinforcing its role as the inactive, photo‐switchable OFF state in this system. The ability to control transport activity with light positions azo‐BUs as promising candidates for the development of next‐generation, stimuli‐responsive anticancer agents. This work introduces a reversible photo‐gated anion channel with therapeutic potential, offering a powerful platform for studying membrane transport and designing light‐responsive biomedical tools.

MViT: A vision transformer with fractal path reordering and dynamic positional encoding

Vision Transformers have demonstrated remarkable performance in image classification and structural modeling; however, fixed patch partitioning and static positional encoding often disrupt spatial continuity, thereby limiting their ability to represent rotated structures and irregular boundary regions. To address these limitations, we propose the Moore-curve Vision Transformer (MViT), a Vision Transformer (ViT) framework based on a recursive Moore curve. The proposed framework comprises three key components. First, a multi-order fractal mapping is employed to optimize patch reordering and enhance the spatial coherence of the token sequence. Second, a 7×7 dynamic partitioning template together with a boundary compensation algorithm jointly optimizes dense structural representation and resolution adaptability. Third, a period-aware positional encoding module integrates fractal periodic parameters with convolutional features to align positional embeddings with the fractal traversal pattern. This design significantly enhances the structural adaptability of the model to complex image layouts. Experimental results show that MViT improves classification accuracy over ViT-B/16 by 0.52% and 0.31% on the CIFAR-100 and ImageNet-21k datasets, respectively, while also achieving noticeable improvements in PSNR and SSIM. Ablation and rotational perturbation experiments further confirm its robustness to rotation and localized focus variations. Moreover, MViT exhibits strong structural compatibility, maintaining stable performance across different Transformer backbones and diverse visual tasks.

Strength analysis of cable tunnels with different embedding depths by using finite element method

Dual‐Function Zinc Modulation Stabilizes Ru Clusters on Spinel Oxide for Efficient and Durable Acid Water Oxidation

Abstract The development of low ruthenium (Ru)‐based anodes with high activity and durability is crucial and still a challenge for proton exchange membrane water electrolysis (PEMWE) systems at a low cost. Here, we synthesized a Ru cluster catalyst (Ru clusters /ZnCo 2 O 4 ) loaded on zinc (Zn)‐doped cobalt oxide spinel, where the presence of Zn atoms transforms the support (ZnCo 2 O 4 ) from a “fragile structure” to a “stable substrate” and indirectly stabilizes the Ru active center by optimizing the electronic environment of the Ru sites through electron transfer meanwhile, thereby achieving a balance between high activity and stability in acidic oxygen evolution reaction (OER). The prepared catalysts with a low overpotential (200 mV) and can operate stably for over 725 h at 10 mA cm −2 (with a decay rate of only 0.143 mV h −1 ). A PEMWE device uses Ru clusters /ZnCo 2 O 4 as the anodes operate stably for over 275 h at 200 mA cm −2 , demonstrating promising application potential. Theoretical calculations and experiments reveal Zn atoms can regulate the support electronic structure concurrently, endowing the catalysts with high activity and long lifetime. This strategy provides a new paradigm for the development of acidic OER catalysts: utilizing inert metals to regulate the supports, achieving dual breakthroughs in activity and stability.

Identification of empagliflozin-related hub genes in atherosclerosis and their correlations with immune infiltration: Network pharmacology and bioinformatics analyses

Background Atherosclerosis (AS) is by far the most frequent underlying cause of atherosclerotic cardiovascular disease. Recently, sodium-glucose cotransporter 2 (SGLT2) inhibitors stand out for their anti-atherosclerotic effects. The present study was conducted to explore the potential genetic and molecular mechanisms of empagliflozin, a selective SGLT2 inhibitor, in preventing AS, and the correlation of empagliflozin-related hub genes with immune cells. Methods In our study, pharmacology platforms were accessed to identify the empagliflozin-related genes (ERGs). AS datasets GSE100927 and GSE43292 were downloaded from the public GEO database to identify AS-differentially expressed genes (AS-DEGs). Furthermore, the empagliflozin-related DEGs (ERDEGs) were obtained by intersecting AS-DEGs and ERGs. ERDEGs were further analyzed for Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. The protein-protein interaction (PPI) network was constructed to screen for hub genes, which were subjected to regulatory network construction, ROC curve plotting, as well as validation analysis, correlation and Friends analysis, and CIBERSORT and ssGSEA analysis. Results A total of 33 genes were identified as ERDEGs, among which the empagliflozin-related hub genes were identified to be IL1B, EGFR, ERBB2, JAK2, SYK, and LGALS3 in AS. Furthermore, in terms of the characteristics of immune cells, there were significant correlations between ERBB2 and CD8 + T cells, IL1B and resting mast cells, LGALS3 and eosinophils, as well as JAK2 and CD56dim NK. Conclusion Our network pharmacology and bioinformatics analyses provide a comprehensive understanding of the potential mechanisms of empagliflozin in AS at the genomic level, with the discovery of significant correlations between the screened hub genes and various immune cell subsets.

Hybrid quantum classical framework for electroencephalogram driven neurological processing in epileptic seizure taxonomy

Seawater Electrosynthesis of Hydrogen Peroxide at Industrial‐level Current Densities Enabled by Pentagonal Defect‐Rich Nanocarbon with Chlorine Doping

Abstract The electrosynthesis of hydrogen peroxide (H 2 O 2 ) via two‐electron oxygen reduction reaction (2e – ORR) in seawater shows great prospects. However, designing an electrocatalyst with high activity and selectivity, resistance to seawater corrosion, and even stable operation at industrial currents (≥300 mA cm −2 ), remains a critical challenge. In this work, we report a pentagonal defect‐rich nanocarbon with chlorine‐doping (Cl‐PDC) by tailoring fullerene (C 60 ) precursors via the molten salt method. The as‐prepared Cl‐PDC catalyst achieves a record H 2 O 2 yield of 74.61 mol g cat −1  h −1 at a current density of 800 mA cm −2 with a nearly 100% Faradaic efficiency, outperforming among all previously reported catalysts in simulated seawater or neutral environments. Remarkably, the Cl‐PDC‐based electrode maintains operational stability over 400 h in simulated seawater, and enables rapid disinfection and pollutant degradation. Theoretical calculations and experimental analysis reveal that the synergy between the intrinsic pentagonal defects and Cl doping modulates the electronic structure of the carbon framework, optimizing *OOH intermediate adsorption, and introduces the localized negative charge to suppress Cl − poisoning at active sites. This work paves the way for sustainable seawater H 2 O 2 production and marine environmental protection.

Gamma irradiation crosslinked fluorescent nanocarbon based biodegradable hydrogel for controlled release of antibiotics

Controlled and sustained antibiotic delivery is critical for combating antimicrobial resistance while minimizing side effects. Herein, a novel biodegradable hydrogel system, synthesized via gamma irradiation, incorporating fluorescent carbon dots (CDs) as multifunctional nano-crosslinkers, has been reported. The CDs, prepared from sustainable bio-precursors, reinforced the polymer network and enhanced the mechanical stability and swelling behavior, while simultaneously serving as intrinsic fluorescent probes for potential real-time monitoring of degradation and drug release. Thorough characterization revealed consistent morphology, adjustable biodegradability, and enhanced rheological characteristics. Drug release investigations demonstrated a diffusion-controlled mechanism, wherein the integration of CD diminished the cumulative antibiotic release from approximately 70% to approximately 40%, thereby facilitating precise regulation of release kinetics. The single-step gamma irradiation method facilitates concurrent crosslinking and sterilization, providing an efficient and scalable production strategy. This study presents a multifunctional hydrogel platform that integrates sustainable nanomaterials, regulated drug administration, and real-time monitoring, thereby facilitating the development of advanced theragnostic systems.

A study of mental health status and its influencing factors in normal weight obesity population

Deactivation of Single‐Atom Catalysts by Nanoparticles

Abstract Single‐atom catalysts (SACs) represent a pinnacle of atomic efficiency and catalytic precision. Their remarkable activity and selectivity arise from isolated, low‐coordinate metal centers that engage directly in bond‐forming events. However, under realistic reaction conditions, SACs are far from static. Increasing evidence reveals that single atoms undergo dynamic evolution over the reaction time. In this perspective, we challenge the conventional dichotomy that views SACs and nanoparticles (NPs) as fundamentally distinct catalytic systems. We propose that NPs, rather than acting as parallel or cooperative catalysts, may function as catalytic poisonants for SACs by trapping active metal atoms. This transformation results in loss of activity, reduced selectivity, and degradation of the catalytic system. Drawing on mechanistic studies, thermodynamic data, and experimental observations across diverse reaction classes, including hydrogenation, oxidation, and cross‐coupling, we show that the aggregation of SACs into NPs is not merely a side process but rather a limitation to their stability and utility. We further outline thermodynamic and kinetic strategies to suppress this deactivation pathway and propose design principles that elevate NP suppression from a synthetic challenge to a foundational criterion in catalyst development. This perspective reframes the SAC–NP relationship as a dynamic continuum and emphasizes the importance of stabilizing isolated active sites in next‐generation catalytic technologies.