Search Results (1,428)

Under the growing demand for processing multimodal and cross-lingual information, traditional retrieval systems have encountered substantial limitations when handling heterogeneous inputs such as images, textual layouts, and multilingual language expressions. To address these challenges, a unified retrieval framework has been proposed, which integrates visual features from images, layout-aware optical character recognition (OCR) text, and bilingual semantic representations in Chinese and English. This framework aims to construct a shared semantic embedding space that mitigates semantic discrepancies across modalities and resolves inconsistencies in cross-lingual mappings. The architecture incorporates three main components: a visual encoder, a structure-aware OCR module, and a multilingual Transformer. Furthermore, a joint contrastive learning loss has been introduced to enhance alignment across both modalities and languages. The proposed method has been evaluated on three core tasks: a single-modality retrieval task from image → OCR, a cross-lingual retrieval task between Chinese and English, and a joint multimodal retrieval task involving image, OCR, and language inputs. Experimental results demonstrate that, in the joint multimodal setting, the proposed model achieved a Precision@10 of 0.693, Recall@10 of 0.684, nDCG@10 of 0.672, and F1@10 of 0.685, substantially outperforming established baselines such as CLIP, LayoutLMv3, and UNITER. Ablation studies revealed that removing either the structure-aware OCR module or the cross-lingual alignment mechanism resulted in a decrease in mean reciprocal rank (MRR) to 0.561, thereby confirming the critical role of these components in reinforcing semantic consistency across modalities. This study highlights the powerful potential of large language models in multimodal semantic fusion and retrieval tasks, providing robust solutions for large-scale semantic understanding and application scenarios in multilingual and multimodal contexts. Full article

Slow-release nitrogen fertilizers enhance crop production and reduce environmental pollution, but their slow nitrogen release may cause insufficient nitrogen supply in the early stages of crop growth. Mixed nitrogen fertilization (MNF), combining slow-release nitrogen fertilizer with urea, is an effective way to increase yield and income and improve nitrogen fertilizer efficiency. This study used urea alone (Urea) and slow-release nitrogen fertilizer alone (C/SRF) as controls and employed meta-analysis and a random forest model to assess MNF effects on crop yield and nitrogen partial factor productivity (PFPN), and to identify key influencing factors. Results showed that compared with urea, MNF increased crop yield by 7.42% and PFPN by 8.20%, with higher improvement rates in Northwest China, regions with an average annual temperature ≤ 20 °C, and elevations of 750–1050 m; in soils with a pH of 5.5–6.5, where 150–240 kg·ha⁻¹ nitrogen with 25–35% content and an 80–100 day release period was applied, and the blending ratio was ≥0.3; and when planting rapeseed, maize, and cotton for 1–2 years. The top three influencing factors were crop type, nitrogen rate, and soil pH. Compared with C/SRF, MNF increased crop yield by 2.44% and had a non-significant increase in PFPN, with higher improvement rates in Northwest China, regions with an average annual temperature ≤ 5 °C, average annual precipitation ≤ 400 mm, and elevations of 300–900 m; in sandy soils with pH > 7.5, where 150–270 kg·ha⁻¹ nitrogen with 25–30% content and a 40–80 day release period was applied, and the blending ratio was 0.4–0.7; and when planting potatoes and rapeseed for 3 years. The top three influencing factors were nitrogen rate, crop type, and average annual precipitation. In conclusion, MNF should comprehensively consider crops, regions, soil, and management. This study provides a scientific basis for optimizing slow-release nitrogen fertilizers and promoting the large-scale application of MNF in farmland. Full article

Peatlands store substantial amounts of carbon (C) in the form of peat, but are increasingly threatened by drought and shrub encroachment under climate warming. However, how peat decomposition and its temperature sensitivity (Q₁₀) vary with depth and plant litter input under these stressors remains poorly understood. We incubated peat from two depths with different degrees of decomposition, either alone or incubated with Sphagnum divinum shoots or Betula ovalifolia leaves, under five temperature levels and two moisture conditions in growth chambers. We found that drought and Betula addition increased CO₂ emissions in both peat layers, while Sphagnum affected only shallow peat. Deep peat alone or with Betula exhibited higher Q₁₀ than pure shallow peat. Drought increased the Q₁₀ of both depths’ peat, but this effect disappeared with fresh litter addition. The CO₂ production rate showed a positive but marginal correlation with microbial biomass carbon, and it displayed a rather similar responsive trend to warming as the microbial metabolism quotient. These results indicate that both deep and dry peat are more sensitive to warming, highlighting the importance of keeping deep peat buried and waterlogged to conserve existing carbon storage. Additionally, they further emphasize the necessity of Sphagnum moss recovery following vascular plant encroachment in restoring carbon sink function in peatlands. Full article

The force ripple of a permanent magnet synchronous linear motor (PMSLM) caused by multi-source disturbances in practical applications seriously restricts its high-precision motion control performance. The traditional single-mechanism model has difficulty fully characterizing the nonlinear disturbance factors, while the data-driven method has real-time limitations. Therefore, this paper proposes a hybrid modeling framework that integrates the physical mechanism and measured data and realizes the dynamic compensation of the force ripple by constructing a collaborative suppression algorithm. At the mechanistic level, based on electromagnetic field theory and the virtual displacement principle, an analytical model of the core disturbance terms such as the cogging effect and the end effect is established. At the data level, the acceleration sensor is used to collect the dynamic response signal in real time, and the data-driven ripple residual model is constructed by combining frequency domain analysis and parameter fitting. In order to verify the effectiveness of the algorithm, a hardware and software experimental platform including a multi-core processor, high-precision current loop controller, real-time data acquisition module, and motion control unit is built to realize the online calculation and closed-loop injection of the hybrid compensation current. Experiments show that the hybrid framework effectively compensates the unmodeled disturbance through the data model while maintaining the physical interpretability of the mechanistic model, which provides a new idea for motor performance optimization under complex working conditions. Full article

Background/Objectives: The clinical potential of antimicrobial peptides (AMPs) against dual threats like antimicrobial resistance (AMR) and cancer is often limited by their high host cell toxicity. Here, we focused on brevinin-2OS (B2OS), a novel peptide from the skin of Odorrana schmackeri with potent haemolytic activity. The objective was to study the structure–activity relationship and optimise the safety via targeted modifications. Methods: A dual-modification strategy involving C-terminal truncation and subsequent N-terminal D-amino acid substitution was employed. The bioactivities and safety profiles of the resulting analogues were evaluated using antimicrobial, haemolysis, and cytotoxicity assays. Result: Removal of the rana box in B2OS(1-22)-NH₂ substantially reduced haemolysis while maintaining bioactivities. Remarkably, the D-leucine substitution in [D-Leu²]B2OS(1-22)-NH₂ displayed a superior HC₅₀ value of 118.1 µM, representing a more than ten-fold improvement compared to its parent peptide (HC₅₀ of 10.44 µM). This optimised analogue also demonstrated faster bactericidal kinetics and enhanced membrane permeabilisation, leading to a greater than 22-fold improvement in its therapeutic index against Gram-positive bacteria. Conclusions: The C-terminal rana box is a primary determinant of toxicity rather than a requirement for activity in the B2OS scaffold. The engineered peptide [D-Leu²]B2OS(1-22)-NH₂ emerges as a promising lead compound, and this dual-modification strategy provides a powerful design principle for developing safer, more effective peptide-based therapeutics. Full article

Plant mitochondrial genomes are characterized by their complex compositions and structures, large genomes, rapid recombination and evolution rates, and frequent intracellular gene transfer events. Centipedegrass, known as “Chinese turfgrass”, is a warm-season turfgrass that exhibits excellent tolerance to both biotic and abiotic stresses. The chloroplast genome, with 139,107 bp, and the mitochondrial genome, with 564,432 bp, were both assembled into a single circular structure. We identified 44 gene transfer events between the chloroplast and mitochondrial genomes. The mitochondrial gene cox1 could serve as a marker for distinguishing accessions found at different altitudes. The unique features of the centipedegrass mitochondrial genome, coupled with the comparative genomic analysis of both chloroplast and mitochondrial genomes, have the potential to enrich the Poaceae database and provide crucial perspectives on plant evolution, energy metabolism, and responses to environmental conditions. The markers developed could facilitate the analysis of the genetic diversity of centipedegrass. Full article

Background: As COVID-19 transitions to an endemic phase, long COVID symptoms remain a significant public health issue affecting both physical and mental health. A notable proportion of college students report symptoms such as fatigue, cough, and brain fog persisting for weeks or months post-infection. Objectives: This study explored the prevalence and contributing factors of long COVID symptoms among both infected and uninfected students in medicine-related departments. Methods: A cross-sectional study was conducted using online self-reported questionnaires completed by 1523 undergraduate and graduate students in medicine-related departments at a medical university. Participants who had tested positive for COVID-19 within the past three months were excluded. The survey assessed long COVID symptoms, with comparisons conducted between infected and uninfected groups. Multivariate logistic regression identified risk factors associated with long COVID symptoms. Results: Of the 1118 participants, 47.5% of those with a prior COVID-19 diagnosis reported long COVID symptoms within the past month. Significant differences between the infected and uninfected groups were observed in physical, cognitive, and psychological health. Logistic regression identified that prior COVID-19 diagnosis had an association with the presence of long COVID symptoms (odds ratio = 1.48, p = 0.024) after adjusted model analysis. Meanwhile, higher anxiety levels (odds ratio = 1.09, p < 0.001) and a BMI ≥ 24 (odds ratio = 4.50, p < 0.01) were identified as significant risk factors for post-infection syndrome among previously infected students. Sex and exercise habits also influenced symptom prevalence. Conclusions: Since late 2023, with those experiencing cumulative infections surpassing half of Taiwan’s population, long COVID symptoms have persisted as a widespread concern affecting both physical and mental health, continuing into 2025. This study underscores critical risk factors and symptom patterns among students in medicine-related departments, reinforcing the urgency of sustained surveillance and targeted interventions to facilitate comprehensive recovery. Full article

Cadmium (Cd) is a highly toxic heavy metal that can greatly affect crops and pose a threat to food security. Plant growth-promoting rhizobacteria (PGPR) are capable of alleviating the harm of Cd to crops. In this research, a Cd-tolerant PGPR strain was isolated and screened from the root nodules of semi-wild soybeans. The strain was identified as Pseudomonas sp. strain KM25 by 16S rRNA. Strain KM25 has strong Cd tolerance and can produce indole-3-acetic acid (IAA) and siderophores, dissolve organic and inorganic phosphorus, and has 1-aminocyclopropane-1-carboxylate (ACC) deaminase activity. Under Cd stress, all growth indicators of soybean seedlings were significantly inhibited. After inoculation with strain KM25, the heavy metal stress of soybeans was effectively alleviated. Compared with the non-inoculated group, its shoot height, shoot and root dry weight, fresh weight, and chlorophyll content were significantly increased. Strain KM25 increased the superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) activities of soybean seedlings, reduced the malondialdehyde (MDA) content, increased the Cd content in the roots of soybeans, and decreased the Cd content in the shoot parts. In addition, inoculation treatment can affect the community structure of endophytic bacteria in the roots of soybeans under Cd stress, increasing the relative abundance of Proteobacteria, Bacteroidetes, Sphingomonas, Rhizobium, and Pseudomonas. This study demonstrates that strain KM25 is capable of significantly reducing the adverse effects of Cd on soybean plants while enhancing their growth. Full article

Polyurethane foam is widely used as a primary filling material in car seats. While it provides good damping and energy absorption, the mechanical properties are complex but play a vital role in vibration attenuation and vehicle ride comfort. This study proposes a comprehensive experimental and analytical method to characterize the visco-hyperelastic properties of seat-grade polyurethane foam. Quasi-static and dynamic compression tests were conducted on foam blocks to obtain load–deflection curves and dynamic stiffness. A visco-hyperelastic material model was developed, where the hyperelastic response was derived via the hereditary integral and difference-stress method, and viscoelastic behavior was captured using a Prony series fitted to dynamic stiffness data. The model was validated using finite element simulations, showing good agreement with experimental results in both static and dynamic conditions. The proposed method enables accurate characterization of the visco-hyperelastic material properties of seat-grade polyurethane foam. Full article

Although age-related changes in the gut microbiome of pigs have been extensively studied, the dynamic patterns of fecal microbiota and SCFAs during the gestation-to-weaning period in sows remain poorly characterized. We aim to characterize the changes in fecal microbiota and SCFAs from pregnancy to weaning, and to investigate their associations with maternal weight gain during gestation. We systematically collected 100 fecal samples at four time points (day 30 of pregnancy (T1), 1–2 days before delivery (T2), day 10 after delivery (T3), and day 21 of weaning stage (T3)), and measured the body weight of sows at T1 (132 kg ± 10.8) and T2 (205 kg ± 12.1). The primary nutrient components of the diets during the gestation and lactation periods are summarized. All fecal samples were subjected to 16S rRNA gene sequencing. We found that a high proportion of crude fiber (bran) is a key feature of the gestation diet, which may affect enterotype shifts and gut microbial composition. Sows fed a high-fiber diet showed significant enrichment of gut microbiota, including genera such as Prevotellaceae_UCG-003, Prevotellaceae_NK3B31_group, and Prevotella_9 during the gestational period (LDA score > 2). Moreover, Eubacterium_coprostanoligenes_group (average relative abundance: 5.5%) and Lachnospiraceae_NK4A136_group (average relative abundance: 2.5%) were the dominant bacteria during the lactation stage. Fecal propionate and butyrate levels were lowest in late gestation, and propionate negatively and acetate positively correlated with body weight change (p < 0.05). Additionally, certain Prevotella taxa were associated with arachidonic acid metabolism and acetate production (p < 0.05). Our study identified key microbial communities across four stages from gestation to weaning and revealed that dietary patterns can shape the sow gut microbiota. Furthermore, we observed significant correlations between SCFAs and body weight change during pregnancy. These findings provide a scientific basis and theoretical support for future strategies aimed at modulating gut microbiota and targeting SCFAs to improve maternal health and productivity throughout the gestation-to-weaning period. Full article

Aiming at the problems of low thermal analysis efficiency and high computational cost of traditional computational fluid dynamics (CFD) methods for modular fault-tolerant permanent magnet synchronous motors (MFT-PMSMs) under complex working conditions, this paper proposes a fast modeling and calculation method of motor temperature field based on a high-efficiency and high-precision network algorithm. In this method, the physical structure of the motor is equivalent to a parameterized network model, and the computational efficiency is significantly improved by model partitioning and Fourth-order Runge Kutta method. The temperature change of the cooling medium is further considered, and the temperature rise change of the motor at different spatial positions is effectively considered. Based on the finite element method (FEM), the space loss distribution under rated, single-phase open circuit and overload conditions is obtained and mapped to the thermal network nodes. Through the transient thermal network solution, the rapid calculation of the temperature rise law of key components such as windings and permanent magnets is realized. The accuracy of the thermal network model was verified by using fluid-structure coupling simulation and prototype test for temperature analysis. This method provides an efficient tool for thermal safety assessment and optimization in the motor fault-tolerant design stage, especially for heat capacity check under extreme conditions and fault modes. Full article

Prolonged high-dose administration of synthetic glucocorticoids (GCs) leads to limb muscle atrophy and weakness, yet its underlying mechanisms remain incompletely understood. Muscle fibers and muscle satellite cells (MSCs) are essential for skeletal muscle development and associated pathologies. This study demonstrates that dexamethasone (Dex) induced MSC-derived extracellular vesicles (EVs) impair myogenesis in muscle fiber-like cells (MFLCs) via inducible nitric oxide synthase (iNOS) suppression. High-throughput sequencing revealed a marked upregulation of miR-335-5p in MSC-derived EVs following Dex treatment. Mechanistically, EV miR-335-5p targeted MAPK11, leading to iNOS downregulation and subsequent UPS activation in MFLCs, which directly promoted muscle protein degradation. Collectively, our findings identify the EV miR-335-5p/MAPK11/iNOS axis as a critical mediator of GC-induced muscle atrophy, offering novel insights into therapeutic strategies targeting EV-mediated signaling in muscle wasting disorders. Full article

Alzheimer’s disease (AD) is a prevalent neurodegenerative disorder with a pressing need for novel therapeutics. However, current medications only offer symptomatic relief, without tackling the underlying pathology. To explore the bioactive potential of marine-derived fungi, this study focused on Aspergillus terreus C23-3, a strain isolated from the coral Pavona cactus in Xuwen County, China, which showed a richer metabolite fingerprint among the three deposited A. terreus strains. AntiSMASH analysis based on complete genome sequencing predicted 68 biosynthetic gene clusters (BGCs) with 7 BGCs synthesizing compounds reported to have anti-AD potential, including benzodiazepines, benzaldehydes, butenolides, and lovastatin. Liquid chromatography coupled with mass spectrometry (LC-MS)-based combinational metabolomic annotation verified most of the compounds predicted by BGCs with the acetylcholinesterase (AChE) inhibitor territrem B characterized from its fermentation extract. Subsequently, molecular docking showed that these compounds, especially aspulvione B1, possessed strong interactions with AD-related targets including AChE, cyclin-dependent kinase 5-p25 complex (CDK5/p25), glycogen synthase kinase-3β (GSK-3β), and monoamine oxidase-B (MAO-B). In conclusion, the genomic–metabolomic analyses and molecular docking indicated that C23-3 is a high-value source strain for anti-AD natural compounds. Full article

Peptides play a crucial role in the development of pharmaceuticals and functional foods. Multiple studies have shown that natural bioactive peptides possess antioxidant, antihypertensive, anti-tumor, and anti-inflammatory activities. Marine bioactive peptides, especially those sourced from fish, constitute a substantial reservoir of these molecules. Although considerable research has been undertaken on fish-derived peptides, studies specifically concerning those from tuna are limited. Tuna, a marine fish of high nutritional value, generates substantial by-product waste during fishing and processing. Therefore, it is essential to conduct an evaluation of the advancements in study on tuna-derived active peptides and to offer a perspective on the direction of future investigations. This review integrates prospective bioactive peptides derived from tuna and reports contemporary strategies for their investigation, including extraction, purification, screening, identification, and activity evaluation procedures, including Yeast Surface Display (YSD) and molecular docking. This review seeks to promote the continued investigation and application of bioactive peptides derived from tuna. Full article

When a high-speed rotating projectile faces high impact loads, the sensitive parts of the control system can get damaged, resulting in operational failure. It is crucial to develop a unique buffer structure that offers impact resistance and has a small contact area. An annularly distributed multi-sphere point contact structure was designed and fabricated on a magnesium alloy substrate based on the Hertz contact theory. The accuracy of the finite element numerical model, constructed using Abaqus/Explicit, was verified through hydraulic impact tests. The impact mechanical properties of the structure were studied by analyzing the influence of the number, diameter, and cavity radius of hemispheres using an experimentally verified finite element model. The axial and radial deformations of the structure were compared and analyzed. The research findings indicate that the deformation and impact resistance of the structure can be greatly influenced by increasing the number of hemispheres, enlarging the hemisphere diameter, and incorporating internal cavities. Specifically, with 6 hemispheres, each with a diameter of Φ 6 mm and a cavity radius of R1.5 mm, the axial and radial deformations are only 1.03 mm and 3.02 mm, respectively. The contact area of a single hemisphere is 7.16 mm². The study offers new perspectives on choosing buffer structures in high-impact environments. Full article