Search Results (106,437)

Financial reporting irregularities and audit modifications represent important warning signals of elevated fraud and financial distress risk. While recent studies report high predictive accuracy in fraud detection, most approaches frame the problem as a purely algorithmic classification task and offer limited interpretability for auditors, regulators, and decision-makers. This study reframes financial statement analysis as a human-interpretable audit modification risk prediction problem. It integrates domain-informed feature engineering with machine learning models. Using firm-level financial data and audit disclosures, audit opinions are used as a proxy indicator of elevated fraud-related reporting risk rather than confirmed fraudulent behavior. Logistic Regression, Random Forest, and Gradient Boosting models are trained under class imbalance using cost-sensitive learning and evaluated with recall, ROC–AUC, precision, F1-score, and accuracy. The results demonstrate that humanized categorical representations preserve predictive performance while substantially enhancing interpretability. Permutation-based feature importance analysis further identifies financially intuitive risk patterns and threshold-like conditions associated with elevated audit modification risk. The findings suggest that interpretable, risk-oriented machine learning frameworks can support more transparent and actionable financial reporting risk monitoring systems. Beyond predictive performance, the study discusses human-centered challenges related to model interpretability, decision support, and the integration of machine-learning systems into real-world financial reporting and audit-risk assessment workflows. Full article

Broccoli is rich in glucosinolates (GSLs), secondary metabolites that contribute to both plant defense and human health. Optimizing the composition of major aliphatic GSLs is an important breeding objective, yet robust molecular markers for marker-assisted selection (MAS) remain limited. In this study, candidate gene-based kompetitive allele-specific PCR (KASP) markers were developed from conserved GSL biosynthesis genes, focusing on AOP2 and GSL-OH selected from 19 GSL-related genes. Marker–trait associations were evaluated in a natural broccoli population and further validated in an independent F₂ population. Among the tested markers, S101, located in AOP2, exhibited consistent genotype-dependent effects on GNA and PRO across both populations, supporting its stable predictive value. Receiver operating characteristic (ROC) analysis further confirmed strong classification performance of S101 for distinguishing high- and low-content genotypes of these traits in the F₂ population. In contrast, S074 and S035 showed population-dependent effects, with significant associations detected only in the natural population. Although association signals were reduced under mixed linear model (MLM) analysis with false discovery rate (FDR) correction, major loci identified under the general linear model (GLM) framework remained detectable. Overall, these results demonstrate the potential of candidate gene-based KASP markers for improving aliphatic GSL composition in broccoli through marker-assisted selection. Full article

Mental health disorders (MHDs) and acute coronary syndromes (ACSs) demonstrate reciprocal pathophysiological connections with substantial prognostic implications. Despite robust evidence linking MHDs to adverse cardiovascular outcomes, the bidirectional relationship remains inadequately characterized in clinical practice, with limited integration of mental health screening into routine cardiac care pathways. The present narrative review comprehensively presents contemporary data on epidemiology, shared biological mechanisms, clinical consequences, and integrated management strategies across the MHD–ACS continuum. A synthesis of peer-reviewed literature, meta-analyses, observational cohorts, randomized trials, and international guideline documents was performed, focusing on depression, anxiety, post-traumatic stress disorder, bipolar disorder, schizophrenia, and suicidality in relation to ACSs. MHDs are highly prevalent in ACS populations and independently predict increased mortality, major adverse cardiac events, and poorer functional recovery. Shared mechanisms include chronic low-grade inflammation, autonomic imbalance, hypothalamic–pituitary–adrenal axis hyperactivation, platelet hyperreactivity, and endothelial dysfunction. Selective serotonin reuptake inhibitors and cognitive behavioral therapy demonstrate the strongest evidence for treating depression in cardiac populations. Collaborative, stepped-care, and integrated cardiac rehabilitation models consistently improve psychological outcomes, with variable effects on cardiovascular endpoints. MHDs and ACSs form a self-reinforcing clinical continuum. Routine mental health screening and integrated cardio-psychiatric care represent essential components of secondary prevention and long-term outcome optimization. Full article

Circulating microRNAs (miRNAs) are promising minimally invasive biomarkers for cancer risk assessment, yet prospective evidence for breast cancer (BC) remains limited. We conducted a nested case–control study within a prospective cohort to examine whether pre-diagnostic circulating miRNAs are associated with subsequent BC risk and to explore their potential relevance in prospective population-based settings. Baseline serum from 160 women (80 incident BC cases; 80 matched controls) was analyzed, with a median time to diagnosis of 8.9 years. Eight candidate miRNAs were quantified by droplet digital PCR (ddPCR) and normalized to miR-484. Group differences were evaluated by non-parametric tests, and odds ratios for BC were estimated using logistic regression models adjusted for established risk factors, with Bonferroni correction for multiple testing. Cases and controls were comparable at baseline. Among the candidates, lower circulating miR-181 levels showed a suggestive inverse association with BC risk in fully adjusted models, while lower Let7 levels showed only a non-significant, hypothesis-generating inverse trend that did not survive Bonferroni correction. No other miRNA displayed clear associations with BC risk. These findings, while preliminary, support further large-scale prospective investigations specifically designed to assess predictive performance and external validation. employing standardized pre-analytical and analytical protocols, repeated sampling, and independent replication/external validation to clarify the etiologic relevance and potential risk-prediction value of circulating miRNAs for BC. Full article

Electromagnetic scattering from electronic platforms degrades system performance, increases radar detectability, and intensifies electromagnetic interference in modern radar and communication systems. Electromagnetic absorbing layers offer an effective approach for radar cross-section (RCS) reduction; however, existing machine-learning-based design methods rely on black-box, composition-specific models lacking physical interpretability and generalizable design rules. In this work, a physics-informed and interpretable machine learning framework is proposed for application-oriented electromagnetic absorber design in electronic systems. Physically meaningful electromagnetic descriptors related to impedance matching and attenuation are embedded into an explainable learning model to establish transparent relationships between absorber parameters and reflection-related performance. Unlike prior approaches, SHAP-based interpretability is applied to extract universal, quantitative design rules, and ML-driven inverse design is explicitly validated for electronic-system-level RCS reduction. Experimental validation confirms that the predicted designs achieve reflection-related performance with deviations below 5%, demonstrating the reliability of the proposed framework. Full article

As artificial intelligence (AI) systems are increasingly deployed in real-world applications, concerns regarding the unauthorized use of copyrighted images during model training have become more pronounced. In particular, both generative and discriminative models may implicitly internalize distinctive visual patterns from copyrighted data, leading to potential ethical and legal risks even after data removal. In this study, we propose a practical copyright protection framework, termed the Predictable-Image Solution (PIS), which aims to disrupt the learning of copyrighted visual features during the training process. PIS leverages Layer-wise Relevance Propagation (LRP) to identify image regions that contribute positively to a model’s prediction and selectively modifies these regions using non-copyrighted visual substitutes, such as textures or benign image patterns. By targeting semantically influential regions rather than applying global perturbations, the proposed approach effectively interferes with feature extraction while preserving the perceptual quality and overall visual structure of the original image. Extensive experiments conducted on multiple pre-trained image classification models demonstrate that PIS consistently degrades classification performance on protected images, while maintaining high visual similarity as measured by perceptual metrics. These results indicate that PIS offers an effective, model-agnostic, and visually unobtrusive solution for mitigating unauthorized exploitation of copyrighted images in practical AI training scenarios. Full article

Real-time monitoring of wire bow is critical for ensuring wafer quality and preventing wire breakage in diamond multi-wire sawing (MWS). However, the deployment physical sensors in industrial MWS environments is hindered by severe sludge contamination, limited installation space, and high maintenance costs. To address these challenges, this paper proposes a novel data-driven soft sensor framework utilizing machine learning methods to predict wire bow based on readily accessible process data. A feature engineering pipeline, combining variance thresholding and correlation analysis, is established to identify key process variables. Subsequently, six representative ML algorithms are systematically evaluated, with eXtreme Gradient Boosting (XGBoost) optimized via two-stage hyperparameter optimization emerging as the superior model. Experimental results from an industrial MWS machine demonstrate that the proposed model achieves a coefficient of determination ($R^2$) of 0.992 and a mean absolute error (MAE) of 0.116 mm. Furthermore, the prediction is also extended to spatially distributed positions (head, middle, and tail) of the wire web. Finally, SHAP (SHapley Additive exPlanations) is utilized to elucidate the mechanical dependencies. This work provides a reliable and low-cost solution for wire bow monitoring during the MWS process. Full article

This study adheres to a logistic regression modeling protocol originally developed for long-range intelligence analyses and employs data from UN demographic estimates (the 2024 revision) to generate a set of statistical functions that suggest a moderately strong relationship between increasing median age and the probability of a persistently positive international net migration rate (NMR). According to this relationship, the post-Cold War probability (data from 1990 to 2015) of experiencing a persistently positive net migration rate (defined as a +NMR, directly followed by five consecutive years of +NMRs) rose from less than 0.12 at a population median age of 15 years, to a probability greater than 0.55 at 36 years, and then to more than 0.77 at 45 years. The author hypothesizes a speculative set of predictions aimed at providing long-term tests for this model. These predictions assume that, by a median age of 36.0 years, at least one country in the hypothesized cluster of countries will have shifted to experiencing a series of +NMRs. If, as this model predicts, the age-structurally associated transition to sustained +NMRs transpires by 2055, there could be a substantially larger pool of migrant net-receiving states in parts of Asia, Latin America, and North Africa than the UN’s future scenarios currently project. Full article

Because dust-emission processes driven by local, small-scale winds (e.g., terrain-induced winds) are difficult to accurately capture with mesoscale or larger-scale predictive models, this study employed a CFD-Lagrangian particle-tracking approach to numerically simulate near-surface dust release and transport under different atmospheric stability conditions in the same local flow field. The novelty of this work was the integration of MOST-based stable/neutral/unstable inflow construction with Lagrangian particle tracking, enabling a consistent comparison of stability effects within one framework. This framework is useful for assessing local blowing-sand impacts on short-range receptors. A near-surface source term was specified for PM10-class mineral dust, and particles were emitted using a vertically exponential allocation. Simulations were conducted over a kilometer-scale flow domain containing an idealized cosine hill, and the low-level concentration patterns and dispersion-height variations in the resulting dust cloud were analyzed. Compared with neutral conditions, stable stratification produced higher near-surface concentrations and a lower dispersion height, whereas unstable stratification yielded lower near-surface concentrations and a higher dispersion height; as the $\left|L\right|$ increased, the unstable cases gradually approached the neutral state. The influence of reference wind speed exhibited clear stability dependence: under stable conditions, stronger winds intensified the buoyancy-related suppression of dust dispersion, while under unstable conditions, stronger winds inhibited the vertical spreading of the dust cloud. In addition, reduced air density representative of plateau environments resulted in lower dust-cloud concentrations and higher dispersion heights. These findings highlight the coupled effects of stratification and wind speed on near-field dust dispersion and provide a reference for assessing local dust emissions over complex terrain. Full article

The aim of this work was to develop a model using Artificial Neural Networks (ANN) to predict stem cutting parameters for giant miscanthus. Experimental studies were conducted to determine biometric traits: maximum stem diameter (Dmax), minimum stem diameter (Dmin), stem wall thickness (THwall), and strength parameters (cutting force, cutting work) for two giant miscanthus genotypes, depending on the internode number (NrNod) and water content (MC). A total of 600 measurement results were obtained, which were randomly divided into training (60%), test (20%), and validation (20%) subsets. Two semantic models were adopted: one for predicting stem cutting force (ann1) and one for predicting cutting work (ann2). The independent variables (ANN inputs) were: Gen, MC, NrNod, Dmax, Dmin, and THwall. The ANN creation process was performed using Statistica Neural Networks. For each of the two semantic models (ANN1 and ANN2), 100 neural networks were developed, with the top 10 ANNs retained for further analysis. The criterion for selecting the best neural network was the root mean square error (RMSE) for the test subset. For ANN1, the RMSE values varied from 6.89 N to 8.70 N. For ANN2, the RMSE values varied from 0.086 J to 0.102 J. For the most accurate ANN1-03 (MLP 7-10-1), used to predict grass cutting force, the RMSE values were 6.46 N–6.89 N–4.70 N for the training, test, and validation subsets. For the most accurate ANN2-02 (MLP 7-10-1), used to predict grass cutting work, the RMSE values were 0.0646 J–0.0857 J–0.0596 J for the training, test, and validation subsets. Full article

Port-area hydrogen refueling stations face low-frequency but high-consequence events when high-pressure leaks ignite as jet fires in wind-exposed, constrained environments. This study develops a consequence-based framework coupling theoretical screening, CFD combustion analysis, and hazard zoning to support separation-distance setting and emergency planning. A jet-fire model estimates flame-impingement distances for multiple leak diameters, and a weighted multi-point radiation model predicts heat-flux fields, from which lethal and irreversible-injury zones are delineated using thresholds of 7 and 5 kW/m², respectively. To move beyond wind-free screening, steady reacting-flow CFD is conducted for a representative release under four ambient conditions, with 4.34 m/s adopted as the representative wind speed for the windy cases based on Ningbo Port conditions. Validation against a visible-flame correlation defined by T ≥ 1573 K shows a deviation of 6.99%. Results show that radiation footprints expand markedly with diameter, with lethal and injury distances scaling approximately linearly within the studied range. Under wind, near-ground hot-plume extents defined by T ≥ 388 K and T ≥ 582 K depend strongly on wind direction and station geometry, whereas visible flame length is less sensitive. Additional sensitivity analyses indicate that the quasi-steady results are weakly affected by the selected ignition snapshot, while inclined releases modify projected plume/flame extents without altering the main engineering interpretation of the baseline case. The results support theory-based preliminary screening, but wind direction should be explicitly considered in exclusion-zone definition. Full article

Acidic chitinase (Chia) degrades chitin, a structural polysaccharide in insect exoskeletons, and plays important roles in omnivorous and insectivorous mammals and birds. In birds, gene duplications have generated multiple Chia paralogs with functional divergence, but the molecular basis for this diversification remains unclear. Here, we characterized three chicken Chia paralogs (Chia1–3) and identified distinct pH-dependent enzymatic profiles. Chia1 is enzymatically inactive but was captured by chitin-affinity resin despite lacking a canonical chitin-binding domain, suggesting residual substrate interaction through the catalytic domain or a non-catalytic role. Chia2 exhibits maximal activity at pH 2.0, whereas Chia3 peaks at pH 5.0 and displays broader activity. Exon swapping and site-directed mutagenesis identified residues 104 (Ala in Chia2, Asp in Chia3) and 269 (His vs. Asn) as key contributors to pH-dependent activity differences. Reciprocal substitutions shifted pH profiles accordingly. Structural modeling and computational pKa predictions suggested that D213 and residue 269 may function as a pKa-regulating module influencing catalytic ionization. Comparative sequence analysis revealed lineage-specific conservation of these residues, consistent with adaptive divergence. Our findings show that limited amino acid substitutions can markedly modify pH-dependent enzymatic activity, providing mechanistic insight into how local residue variation contributes to the functional diversification of duplicated genes. Full article

With the development of 6G communication technology, Space–Air–Ground Integrated Networks (SAGINs) have become critical infrastructure for global intelligent collaborative computing. However, federated learning deployment in SAGINs faces three severe challenges: the high dynamics of satellite orbital motion, node resource heterogeneity, and privacy vulnerabilities in data transmission. This paper proposes an adaptive compressed sensing differential privacy federated learning framework based on orbital spatiotemporal characteristics. First, we design orbital periodicity-driven time-varying sparse sensing matrices that dynamically adjust compression strategies according to satellite orbital positions, achieving intelligent communication efficiency optimization. Second, we propose an orbital predictability-based privacy budget temporal allocation mechanism and perform differential privacy noise injection in the compressed domain, establishing a compression–privacy joint optimization algorithm. Furthermore, we construct an energy–communication–privacy ternary collaborative mechanism that achieves multi-objective dynamic balance through model predictive control. Finally, we design reinforcement learning-based dynamic routing scheduling and hierarchical aggregation strategies to effectively handle the time-varying characteristics of network topology. Simulation experiments demonstrate that compared to existing methods, the proposed approach achieves 3–12% improvement in model accuracy and 30–50% enhancement in communication efficiency while maintaining differential privacy protection with dynamic privacy budget ε ∈ [0.1,10.0]  and compression ratio ρ ∈ [0.2,0.8]. Unlike static compressed sensing approaches that ignore orbital periodicity, the proposed orbital-driven time-varying sensing matrices reduce reconstruction error by up to 19.4% compared to fixed-matrix baselines, validating the synergistic effectiveness of integrating orbital spatiotemporal characteristics with federated learning in 6G SAGIN deployments. The framework assumes reliable orbital propagation via SGP4/SDP4 models and does not account for Doppler frequency shifts or inter-satellite link handover delays; future extensions include scalability to mega-constellations and integration of quantum-resistant privacy mechanisms. Full article

Precise water quality forecasting is vital for sustainable resource management and public health, especially in semi-arid environments. This study investigates the predictive capabilities of ten Machine Learning (ML) algorithms using a dataset of 308 drinking water samples collected from various districts in Şanlıurfa Province, Türkiye. We evaluated ten predictive models, including Support Vector Regressor (SVR) and Extreme Gradient Boosting (XGBoost), both integrated with dimensionality reduction and hyperparameter optimization. Nineteen physicochemical and microbiological parameters—Temperature, chlorine (Cl⁻), pH, Electrical Conductivity (EC), Total Dissolved Solids (TDS), nitrite (NO₂⁻), nitrate (NO₃⁻), ammonium (NH₄⁺), sulfate (SO₄²⁻), Free Chlorine (Cl₂), calcium (Ca²⁺), magnesium (Mg²⁺), sodium (Na⁺), potassium (K⁺), fluoride (F⁻), trihalomethanes (THMs), Escherichia coli, Enterococci, Total Coliform—were used as input features. The dataset was split into training (75%) and testing (25%) subsets, and model performance was assessed through 10-fold cross-validation and hold-out testing procedures. To improve model generalization and mitigate the effects of class imbalance, we implemented the Adaptive Synthetic Sampling (ADASYN) technique. ML algorithms were evaluated using standard regression metrics: Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and the Coefficient of Determination (R²). The LSTM model optimized using Randomized Search outperformed the SVR and XGBoost models, demonstrating the highest accuracy and generalization capability, as evidenced by the superior R² value of 0.999 following ADASYN balancing and the lowest RMSE (1.206). These findings underscore the effectiveness of the LSTM framework in modeling the complex variance of the Weighted Arithmetic Water Quality Index (WAWQI). The findings of this study are expected to support future water quality monitoring strategies, inform policy development, and contribute to sustainable water resource management in arid and semi-arid regions. Full article

The valorization of agro-industrial by-products is a sustainable approach to recovering high-value bioactive compounds. In this study, Opuntia ficus-indica (L.) Mill. seed press residues were investigated as a source of phenolic and flavonoid compounds using microwave-assisted extraction (MAE). A multi-step optimization strategy was implemented, combining preliminary single-factor experiments (OVAT), response surface methodology based on a Box–Behnken design (BBD), and machine learning modeling using K-nearest neighbors coupled with the dragonfly algorithm (KNN_DA), followed by desirability-based validation. The effects of ethanol concentration (50–100%), microwave power (400–800 W), extraction time (2–4 min), and liquid-to-solid ratio (30–50 mL/g) were evaluated on Folin–Ciocalteu reducing capacity (FCRC), AlCl₃ complexation response, and antioxidant activity assessed by DPPH radical scavenging and reducing power assays. Optimal conditions were identified at 50% ethanol, 800 W microwave power, 4 min extraction time, and a liquid-to-solid ratio of 47.28 mL/g. Under these conditions, FCRC reached 376.85 ± 0.23 mg GAE/100 g DW and 49.16 ± 0.33 mg QE/100 g DW for AlCl₃ complexation response, with prediction errors of 2.80% and 0.82%, respectively. The optimized extracts exhibited enhanced antioxidant activity. These findings confirm MAE as a rapid and environmentally friendly technique and highlight the predictive performance of the KNN_DA model for process optimization. Full article