Search Results (829)

This paper selects 50 listed commercial banks in China from 2012 to 2023 as research samples, and employs the fixed effects model and Hansen’s threshold regression method to systematically examine the impact mechanism and non-linear characteristics of FinTech development on the profitability of commercial banks. The key findings are summarized as follows: (1) FinTech significantly undermines the overall profitability of commercial banks by reshaping the competitive landscape of the industry and intensifying the technology substitution effect. This is primarily reflected in the reduction in traditional interest income and the erosion of market share in intermediary business. (2) Heterogeneity analysis indicates that large state-owned banks and joint-stock banks experience more pronounced negative impacts compared to small and medium-sized banks. (3) Additional research findings reveal a significant single-threshold effect between FinTech and bank profitability, with a critical value of 4.169. When the development level of FinTech surpasses this threshold, its inhibitory effect diminishes substantially, suggesting that after achieving a certain degree of technological integration, commercial banks may partially alleviate external competitive pressures through synergistic effects. This study offers crucial empirical evidence and theoretical support for commercial banks to develop differentiated technology strategies and for regulatory authorities to design dynamically adaptable policy frameworks. Full article

In clinical decision support systems (CDSSs), where accurate classification of drug–drug interactions (DDIs) can directly affect treatment safety and outcomes, identifying drug interactions is a major challenge, introducing a scalable approach for classifying DDIs utilizing a finely-tuned biomedical language model. The method shown here uses BiomedBERT, a domain-specific version of bidirectional encoder representations from transformers (BERT) that was pre-trained on biomedical literature, to reduce the number of resources needed during fine-tuning. Low-rank adaptation (LoRA) was used to fine-tune the model on the DrugBank dataset. The objective was to classify DDIs into two clinically distinct categories, that is, synergistic and antagonistic interactions. A pseudo-labeling strategy was created to deal with the problem of not having enough labeled data. A curated ground-truth dataset was constructed using polarity-labeled interaction entries from DrugComb and verified DrugBank antagonism pairs. The fine-tuned model is used to figure out what kinds of interactions there are in the rest of the unlabeled data. A checkpointing system saves predictions and confidence scores in small pieces, which means that the process can be continued and is not affected by system crashes. The framework is designed to log every prediction it makes, allowing results to be refined later, either manually or through automated updates, without discarding low-confidence cases, as traditional threshold-based methods often do. The method keeps a record of every output it generates, making it easier to revisit earlier predictions, either by experts or with improved tools, without depending on preset confidence cutoffs. It was built with efficiency in mind, so it can handle large amounts of biomedical text without heavy computational demands. Rather than focusing on model novelty, this research demonstrates how existing biomedical transformers can be adapted to polarity-aware DDI classification with minimal computational overhead, emphasizing deployment feasibility and clinical relevance. Full article

Acacia mearnsii De Wild. has been introduced to over 150 countries for its economic value. However, it easily escapes from plantations and establishes monospecific stands across plains, hills, valleys, and riparian habitats, including protected areas such as national parks and forest reserves. Due to its negative ecological impact, A. mearnsii has been listed among the world’s 100 worst invasive alien species. This species exhibits rapid stem growth in its sapling stage and reaches reproductive maturity early. It produces a large quantity of long-lived seeds, establishing a substantial seed bank. A. mearnsii can grow in different environmental conditions and tolerates various adverse conditions, such as low temperatures and drought. Its invasive populations are unlikely to be seriously damaged by herbivores and pathogens. Additionally, A. mearnsii exhibits allelopathic activity, though its ecological significance remains unclear. These characteristics of A. mearnsii may contribute to its expansion in introduced ranges. The presence of A. mearnsii affects abiotic processes in ecosystems by reducing water availability, increasing the risk of soil erosion and flooding, altering soil chemical composition, and obstructing solar light irradiation. The invasion negatively affects biotic processes as well, reducing the diversity and abundance of native plants and arthropods, including protective species. Eradicating invasive populations of A. mearnsii requires an integrated, long-term management approach based on an understanding of its invasive mechanisms. Early detection of invasive populations and the promotion of public awareness about their impact are also important. More attention must be given to its invasive traits because it easily escapes from cultivation. Full article

In this paper, we study the classification problem of short-time-window steady-state visual evoked potentials (SSVEPs) and propose a novel deep convolutional network named EEGResNet based on the idea of residual connection to further improve the classification performance. Since the frequency-domain features extracted from short-time-window signals are difficult to distinguish, the EEGResNet starts from the filter bank (FB)-based feature extraction module in the time domain. The FB designed in this paper is composed of four sixth-order Butterworth filters with different bandpass ranges, and the four bandwidths are 19–50 Hz, 14–38 Hz, 9–26 Hz, and 3–14 Hz, respectively. Then, the extracted four feature tensors with the same shape are directly aggregated together. Furthermore, the aggregated features are further learned by a six-layer convolutional neural network with residual connections. Finally, the network output is generated through an adaptive fully connected layer. To prove the effectiveness and superiority of our designed EEGResNet, necessary experiments and comparisons are conducted over two large public datasets. To further verify the application potential of the trained network, a virtual simulation of brain computer interface (BCI) based quadrotor control is presented through V-REP. Full article

This study analyzed the heavy metal tolerance and chromium reduction and the potential of plant growth to promote Rhizobium sp. OS-1. By genetic makeup, the Rhizobium strain is nitrogen-fixing and phosphate-solubilizing in metal-contaminated agricultural soil. Among the Rhizobium group, bacterial strain OS-1 showed a significant tolerance to heavy metals, particularly chromium (900 µg/mL), zinc (700 µg/mL), and copper. In the initial investigation, the bacteria strains were morphologically short-rod, Gram-negative, appeared as light pink colonies on media plates, and were biochemically positive for catalase reaction and the ability to ferment glucose, sucrose, and mannitol. Further, bacterial genomic DNA was isolated and amplified with the 16SrRNA gene and sequencing; the obtained 16S rRNA sequence achieved accession no. HE663761.1 from the NCBI GenBank, and it was confirmed that the strain belongs to the Rhizobium genus by phylogenetic analysis. The strain’s performance was best for high hexavalent chromium [Cr(VI)] reduction at 7–8 pH and a temperature of 30 °C, resulting in a total decrease in 96 h. Additionally, the adsorption isotherm Freundlich and Langmuir models fit best for this study, revealing a large biosorption capacity, with Cr(VI) having the highest affinity. Further bacterial chromium reduction was confirmed by an enzymatic test of nitro reductase and chromate reductase activity in bacterial extract. Further, from the metal biosorption study, an Artificial Neural Network (ANN) model was built to assess the metal reduction capability, considering the variables of pH, temperature, incubation duration, and initial metal concentration. The model attained an excellent expected accuracy (R² > 0.90). With these features, this bacterial strain is excellent for bioremediation and use for industrial purposes and agricultural sustainability in metal-contaminated agricultural fields. Full article

Background/Objectives: A growing body of literature has demonstrated that living with food insecurity (FI) increases risk for mental health concerns in addition to nutritional deficits (e.g., suboptimal micronutrient consumption, excessive macronutrient consumption, malnutrition). Yet, research is needed to improve our understanding of subpopulations potentially at increased risk for mental health concerns when living in the context of FI. The current study examined psychosocial health across women of different developmental life stages all living with FI. Methods: Female clients of a large, urban food bank (N = 680) living with FI completed measures of mental health and health-related quality of life (HRQOL) in a cross-sectional design conducted on site at the food bank. Results: Consistent with past research, FI severity was correlated with poorer psychosocial health across all variables. A multivariate analysis of covariance compared women living with FI across 4 developmental life stages (young adult, early midlife, late midlife, and older adult; age range = 18–94 years), controlling for FI severity and race/ethnicity, on outcomes related to mental health and HRQOL. Women in early and late midlife reported higher anxiety, eating disorder symptoms, and eating-related psychosocial impairment than younger and older women. Conclusions: The mental health toll of living with FI is profound; midlife may comprise a developmental period of increased vulnerability to experience this mental health burden of living with FI for women. Thus, efforts are needed to develop innovative pathways for interventions to support the mental health of midlife women living with FI, likely involving multi-level and/or multicomponent approaches to resource access. Full article

Although cranial growth has been extensively explored, forensic and biological anthropology lack a formal incorporation of how cranial growth processes impact the adult phenotype and downstream biological profile estimations. Objectives: This research uses an ontogenetic framework to identify when interlandmark distances (ILDs) stabilize during growth to reach adult levels of variation and to evaluate patterns of cranial sexual size dimorphism. Methods: Multivariate adaptive regression splines (MARS) were conducted on standardized cranial ILDs for 595 individuals from the Subadult Virtual Anthropology Database (SVAD) and the Forensic Data Bank (FDB) aged between birth and 25 years. Cross-Validated R-squared (CVRSq) values evaluated ILD variation explained by age while knot placements identified meaningful changes in ILD growth trajectories. Results: Results reveal the ages at which males and females reach craniometric maturity across splanchnocranium, neurocranium, basicranium and cross-regional ILDs. Changes in growth patterns observed here largely align with growth milestones of integrated soft tissue and skeletal structures as well as developmental milestones like puberty. Conclusions: Our findings highlight the variability in growth by sex and cranial region and move forensic anthropologists towards recognizing cranial growth as a mosaic, continuous process with overlap between subadults and adults rather than consistently approaching subadult and adult research separately. Full article

Customer churn prediction has become crucial for businesses, yet it poses significant challenges regarding privacy preservation and prediction accuracy. In this paper, we address two fundamental questions: (1) How can customer churn be effectively predicted while ensuring robust privacy protection of sensitive data? (2) How can large language models enhance churn prediction accuracy while maintaining data privacy? To address these questions, we propose FairRAG, a robust architecture that combines differential privacy, retrieval-augmented generation, and LLMs. Our approach leverages OPT-125M as the core language model along with a sentence transformer for semantic similarity matching while incorporating differential privacy mechanisms to generate synthetic training data. We evaluate FairRAG on two diverse datasets: Bank Churn and Telco Churn. The results demonstrate significant improvements over both traditional machine learning approaches and standalone LLMs, achieving accuracy improvements of up to 11% on the Bank Churn dataset and 12% on the Telco Churn dataset. These improvements were maintained when using differentially private synthetic data, thus indicating robust privacy and accuracy trade-offs. Full article

Background: Deep surgical site infections (SSIs) represent a serious complication following abdominal trauma surgery; however, comprehensive risk factor analysis in large trauma populations remains limited. Although surgical site infections are recognized as preventable complications, little is known about the specific risk factors and clinical outcomes associated with deep SSIs in trauma patients at the national level. Methods: A retrospective cohort study analyzed data from the National Trauma Data Bank from 2020–2022, including 1,198,262 trauma patients with complete demographic, injury severity, and surgical procedure data. Deep SSI development, length of hospital stay, intensive care unit utilization, duration of mechanical ventilation, discharge disposition, and in-hospital mortality were assessed. Multivariate logistic regression was used to identify independent risk factors and quantify associations between patient characteristics and deep SSI occurrence. Results: Deep SSIs occurred in 601 patients (0.05%). Affected patients were younger (median 41 vs. 54 years, p < 0.001), predominantly male (73.7% vs. 61.8%, p < 0.001), and exhibited higher injury severity scores (median 17.0 vs. 5.0, p < 0.001). Major abdominal surgery was the strongest independent predictor (OR 3.08, 95% CI: 2.21–4.23, p < 0.001), followed by injury severity score (OR 1.05, 95% CI: 1.04–1.06, p < 0.001) and ICU length of stay (OR 1.04 per day, 95% CI: 1.03–1.05, p < 0.001). Patients with deep SSIs demonstrated dramatically increased hospital stays (89.5% vs. 4.5% exceeding 21 days, p < 0.001), reduced home discharge rates (28.5% vs. 48.9%, p < 0.001), and higher mortality (4.2% vs. 1.2%, p < 0.001). Conclusions: Major abdominal surgery and injury severity are primary risk factors for deep SSIs in trauma patients, with profound impacts on clinical outcomes and healthcare resource utilization. These findings highlight the importance of targeted prevention strategies for high-risk trauma patients undergoing major abdominal procedures and emphasize the significant burden that deep SSIs place on healthcare systems. Full article

The transition towards a power system characterized by a reduced presence of synchronous generators (SGs) and an increased reliance on inverter-based resources (IBRs), including wind, solar photovoltaics (PV), and battery storage, presents new operational challenges, particularly when these sources exceed 50–60% of the system’s demand. While current grid-following (GFL) IBRs, which are equipped with fast and rigid control systems, continue to dominate the inverter landscape, there has been a notable surge in research focused on grid-forming (GFM) inverters in recent years. This study conducts a comparative analysis of the practicality and control methodologies of GFM inverters relative to traditional GFL inverters from a system planning perspective. A comprehensive framework aimed at assisting system developers and consulting engineers in the grid-integration of wide-scale renewable energy sources (RESs), incorporating strategies for the deployment of inverters, capacitor banks, and harmonic filters, is proposed in this paper. The discussion includes an examination of the reactive power capabilities of the plant’s inverters and the provision of additional reactive power to ensure compliance with grid interconnection standards. Furthermore, the paper outlines a practical approach to assess the necessity for enhanced filtering measures to mitigate potential resonant conditions and achieve harmonic compliance at the installation site. The objective of this work is to offer useful guidelines and insights for the effective addition of RES into contemporary power systems. Full article

Beans (Phaseolus vulgaris and P. lunatus) are the major hosts of bean golden mosaic begomovirus (BGMV). Robust taxonomic criteria were established for Begomovirus species demarcation. However, DNA–A identities among BGMV isolates display a continuous variation (89–100%), which conflicts with the current concept of a single viral species. The diversity of 146 Brazilian isolates designated in the GenBank as BGMV was assessed by comparing their complete DNA–A sequences. The isolates were clustered into four groups, being discriminated mainly by their original Fabaceae hosts. Additional Sequence Demarcation Tool analyses indicated that BGMV-related viruses comprise two clear-cut groups: isolates reported infecting mainly P. vulgaris (identities of 96–97% to the reference NC_004042 isolate) and a group associated with P. lunatus (identities of 89–91%). Moreover, we recognized a distinct set of genomic features in the iterons and Rep-associated protein motifs across these two diversity groups. The host prevalence and genomic differences suggest that most P. lunatus isolates are currently misclassified as BGMV strains, being more likely samples of a closely related (but distinct) Begomovirus species. Hence, the implications of this BGMV diversity should be taken into consideration by classical and biotech breeding programs aiming for large-spectrum viral resistance in Phaseolus species. Full article

Rainfall–runoff processes and flood propagation were quantified to clarify floodwater dynamics in the upper Tenryu River basin. The basin is characterized by contrasting runoff behaviors between its left- and right-bank subbasins and large upstream river storage created by gorge topography. Radar rainfall and dam inflow data were analyzed to determine the runoff characteristics, on which the rainfall–runoff simulation was based. A higher storage capacity was observed in the left-bank subbasins, while an exceptionally large specific discharge was observed in one of the right-bank subbasins after several hours of intense rainfall. Based on these findings, the basin-scale storage was quantitatively evaluated. Water level peaks in the main channel appeared earlier at downstream locations, indicating that tributary inflows strongly affect the flood peak timing. A two-dimensional unsteady model successfully reproduced this behavior and captured the delay in the flood wave speed due to the complex morphology of the Tenryu River. The average α value, representing the ratio of flood wave speed to flow velocity, was 1.38 over the 70 km study reach. This analysis enabled quantification of river channel storage and clarified its relative relationship to basin storage, showing that river channel storage is approximately 12% of basin storage. Full article

(1) Background: Hemorrhagic fever with renal syndrome (HFRS) remains a prevalent zoonosis in Eurasia. Orthohantavirus puumalaense (PUUV), carried by bank voles (Myodes glareolus), is the principal zoonotic pathogen of HFRS in this region. Despite ongoing efforts to develop effective drugs and vaccines against PUUV, this challenge remains. (2) Aim: In this study, we aimed to express a large quantity of the PUUV recombinant N (rN) protein using E. coli. We also sought to develop a protocol for extracting the rN protein from pellets, solubilizing, and refolding it to restore its native form. This protocol is crucial for producing a large quantity of rN protein to develop vaccines and diagnostic tools for HFRS. (3) Methods; PUUV S segment open reading frame (ORF) coding for N protein was synthesized and cloned into the plasmid vector pET-28 (A+). The ORF was transformed, expressed and induced in BL21(DE3) pLysS E. coli strain. Subsequently, rN protein was purified using immobilized metal affinity and ion chromatography. Immune reactivity of rN protein was tested by employing in house and commercial VektoHanta-IgG kit ELISA methods (both in vitro and in vivo). (4) Results: The best conditions for scaling up the expression of the PUUV rN protein were an incubation temperature of 20 °C during a 20 h incubation period, followed by induction with 0.5 mM IPTG. The most significant protein yield was achieved when the pellets were incubated in denaturing buffer with 8M urea. The highest yield of refolded proteins was attained using non-denaturing buffer (50 mM Tris-HCl) supplemented with arginine. A final 50 μL of PUUV rN protein solution with a concentration of 7 mg/mL was recovered from 1 L of culture. The rN protein elicited an antibody response in vivo and reacted with serum taken from patients with HFRS by ELISA in vitro. (5) Conclusion: Therefore, the orthohantavirus N protein’s ability to elicit immune response in vivo suggests that it can be used to develop vaccines against PUUV after conducting in vitro and in vivo studies to ascertain neutralising antibodies. Full article

This paper suggests an extensive inferential method for the Power Pratibha Distribution (PPD) under Unified Hybrid Censoring Schemes (UHCSs), since there is a growing interest in flexible models in both reliability and service operations. This work studies the PPD model using standard Maximum Likelihood Estimation methods and modern Bayesian approaches too. Using a complex architecture, UHCS simulates tests more closely to what is done in practice than by using more basic censoring schemes. Using analysis, the probability and statistical ranges are carefully calculated for the parameters. Tests demonstrate that Bayesian estimation gives better results than many other methods for estimation, especially when the dataset is not very large and when a lot of data is missing. Real-world tests of electromigration failure data and banking service times help to test the methods. In both situations, the PPD shows it can be used successfully in different reliability settings. By joining advanced censoring models and reliable statistical methods, this research gives a helpful toolset to experts in reliability analysis and statistics. Full article

Large-scale land restoration projects require on-the-ground monitoring and evidence-based evaluation. This study, part of the World Bank Burundi Landscape Restoration and Resilience Project (in French: Projet de Restauration et de Résilience du Paysage du Burundi-PRRPB), examines the impact of slow-forming terraces on surface conditions and erosion in Isare (Mumirwa) and Buhinyuza (Eastern Depressions), Burundi. Slow-forming, or progressive, terraces were installed on 16 December 2022 (Isare) and 30 December 2022 (Buhinyuza), featuring ditches and soil bunds to enhance soil and water conservation. Twelve plots were established, with 132 measurement pins, of which 72 were in non-terraced plots (n_PT) and 60 were in terraced plots (PT). Monthly measurements, conducted until May 2023, assessed erosion reduction, surface conditions, roughness, and soil thickness. Terracing reduced soil loss by 54% in Isare and 9% in Buhinyuza, though sediment accumulation in ditches was excessive, especially in n_PT. Anti-erosion ditches improved surface stability by reducing slope length, lowering erosion and runoff. Covered Surface (CoS%) exceeded 95%, while Opened Surface (OS%) and Bare Surface (BS%) declined significantly. At Isare, OS% dropped from 97% to 80%, and BS% from 96% to 3% in PT. Similar trends appeared in Buhinyuza. Findings highlight PRRPB effectiveness in this short-term timeframe, and provide insights for soil conservation in steep-slope regions of Central Africa. Full article