To investigate the proteomic signatures of heart failure (HF) in diabetes. The underlying mechanisms of the elevated risk of HF in diabetes are unknown. In 10 189 ARIC study (Atherosclerosis Risk in Communities) participants free of HF (mean age 57±7 years, 56% women, 22% Black adults, 14% with diabetes), we conducted discovery and internal validation for the associations of 4955 plasma proteins with HF by diabetes status. We performed (1) Cox regression to identify proteins associated with HF by diabetes status, (2) external validation in the MESA study (Multi-Ethnic Study of Atherosclerosis, n=5233, 633 with diabetes), and (3) pathway analyses for identified proteins. Over 24 years in ARIC, there were 2417 HF events (605 among individuals with diabetes). In 993 individuals with diabetes in the discovery sample, 19 proteins were associated with HF (P<10-5), 12 proteins replicated in the internal validation sample (P<0.05/19). Six of the internally validated proteins replicated in MESA (false discovery rate, q<0.05). Five proteins were specifically associated with HF in those with diabetes: 4 are novel (inactive tyrosine-protein kinase 7, chondroadherin, leucine-rich repeat, and immunoglobulin-like domain-containing nogo receptor-interacting protein 1 and fibulin-5) and 1 is the previously known (cartilage intermediate layer protein 2). NPPB (N-terminal pro-BNP) was associated with HF in those with and without diabetes. Pathways over-represented among proteins associated with diabetes-related HF were lipid metabolism, inflammation, and brown adipose tissue (false discovery rate, q<0.05). We identified 5 proteomic markers (4 novel) uniquely related to HF risk among individuals with diabetes and not among those without diabetes.

Initially characterized as a component of the extracellular matrix (ECM) in cartilage, cartilage intermediate layer protein 2 (CILP2) is now recognized as a pleiotropic secretory protein with far-reaching roles in physiology and disease. This review synthesizes evidence establishing CILP2 as a key modulator at the nexus of metabolic dysfunction, cancer, and other pathologies. Genomic studies have firmly established the NCAN-CILP2 locus as a hotspot for genetic variants influencing dyslipidemia and cardiovascular risk. Functionally, CILP2 is upregulated by metabolic stress, including high glucose and oxidatively modified LDL (oxLDL), and actively contributes to pathologies such as dyslipidemia, diabetes, and sarcopenia by impairing glucose metabolism and mitochondrial function. Its role extends to fibrosis and neurodevelopment, promoting hypertrophic scar formation and neurogenesis through interactions with ATP citrate lyase (ACLY) and Wnt3a, respectively. More recently, CILP2 has emerged as an oncoprotein, overexpressed in multiple cancers, including pancreatic ductal adenocarcinoma and colorectal cancer. It drives tumor proliferation and metastasis and correlates with tumor microenvironment remodeling through mechanisms involving Akt/EMT signaling and immune infiltration. The dysregulation of CILP2 in patient serum and its correlation with disease severity and poor prognosis highlight it as a promising biomarker and a compelling therapeutic target across a spectrum of human diseases.

Pyrethroid pesticide residues pose a significant global public health challenge, particularly in complex edible fungus matrices where trace, structurally similar pesticides are difficult to distinguish and detect. To address this critical gap, a novel surface-enhanced Raman spectroscopy (SERS) platform is presented integrating "magnetic enrichment, covalent organic framework (COF)-mediated capture, silver shell enhancement, and machine learning (ML)-driven intelligent recognition". The core innovation lies in the tailored Fe₃O₄@COF@Ag nanocomposite architecture where the magnetic Fe₃O₄ core enables rapid sample separation, the COF intermediate layer provides specific molecular capture, and the Ag shell generates dense SERS "hotspots",synergistically enhanced by ML algorithms to enable precise discrimination of structurally analogous pesticides. The developed Fe₃O₄@COF@Ag substrate demonstrated exceptional performance, achieving detection limits as low as 5.21µg/kg for deltamethrin (CF), 6.08µg/kg for fenvalerate (FV), and 11.6µg/kg for lambda-cyhalothrin (LCF). The method exhibited outstanding linearity (R² > 0.99) for quantitative analysis and anti-interference capability, with recoveries of 94.36-106.89% in real samples. By integrating ML algorithms, the platform achieved 100% classification accuracy for distinguishing structurally similar pyrethroids. Principal component analysis (PCA) and support vector machine (SVM) models effectively extracted discriminative spectral features, enabling precise identification even at trace concentrations. This work provides a rapid, sensitive, and field-deployable solution for pesticide monitoring, offering significant potential to enhance food safety and public health protection.

Abstract Sb 2 Se 3 is an earth-abundant, non-toxic, and low-cost photovoltaic material with promising optoelectronic properties. With a direct bandgap of about 1.2 eV, Sb 2 Se 3 is well suited as the bottom cell in tandem architectures, extending spectral coverage and improving power conversion efficiency. In this study, Silvaco TCAD was employed to simulate solar cells based on inorganic CsPbI 3 (bandgap approximately 1.7 eV) and Sb 2 Se 3 . To improve the performance of Sb 2 Se 3 solar cells, the effect of MoO 3 as a hole transport layer was investigated. In the two-terminal tandem devices, we systematically investigated the effects of defect states, doping concentrations, and intermediate recombination layer thickness on the performance of the CsPbI 3 top cell and Sb 2 Se 3 bottom cell. We also examined the influence of the thicknesses of the CsPbI 3 and Sb 2 Se 3 absorber layers on the overall performance of the tandem device. Current matching can be achieved when the thickness of the CsPbI 3 layer is 259 nm and that of the Sb 2 Se 3 layer is 1000 nm. The device exhibits a current density of 17.45 mA/cm 2 , an open-circuit voltage of 2.139 V, a fill factor of 87.57%, and a power conversion efficiency of 32.66%.

Mitigating vertical clogging and managing blockage migration in bioelectrochemical constructed wetlands: Electrification time regulation as a strategy.

Although traditional homogeneous SiCp reinforced aluminum matrix composites have characteristics such as high specific strength, low thermal expansion coefficient, and excellent wear resistance, but their isotropic properties are difficult to meet the gradient requirements of material properties under complex working conditions. Therefore, SiCp reinforced aluminum matrix composite gradient materials have attracted much attention in aerospace, defense and military industries. This study focuses on the preparation of SiCp/6092Al composite materials with different silicon carbide contents (15%, 20%, and 25%) using the powder metallurgy method. Gradient composite materials were prepared using the ECAP method, and the resulting variations in their microstructure and properties were systematically analyzed. The results indicate that: Based on powder metallurgy technology and large plastic deformation ECAP technology, SiCp/6092Al gradient composite materials with good interfacial bonding have been prepared, and the results of hardness and tensile strength tests show that compared with single volume fraction materials, the prepared gradient composite material has the characteristics of surface ablation resistance, intermediate layer high thermal conductivity, and matrix toughening.

In this paper, we present a Neuron Abandoning Attention Flow (NAFlow) method to address the unsolved problem of visually explaining the attention evolution dynamics inside CNNs when making their classification decisions. A novel cascading neuron abandoning back- propagation algorithm is designed to precisely exclude the abandoned neurons on all intermediate layers inside a CNN model for the first time. Firstly, a Neuron Abandoning Back-Propagation module is proposed to generate Back-Propagation Feature Maps (BPFM) by using inverse function of the intermediate layers of CNN models, on which the neurons not used for decision-making are removed. Meanwhile, the cascading NA-BP modules calculate the tensors of importance coefficients which are linearly combined with the tensors of BPFMs to form the NAFlow. Secondly, to be able to visualize attention flow for similarity metric-based CNN models, a new channel contribution weights module is proposed to calculate the importance coefficients via Jacobian Matrix. Extensive evaluations demonstrate the effectiveness of the proposed NAFlow across eleven widely-used CNN models for various tasks of general image classification, contrastive learning classification, few-shot image classification, and image retrieval.

AISI 1045 steel often undergoes premature failure under combined corrosive-wear conditions due to its insufficient surface durability. To address this, a hot-dip aluminum (HDA) coating was deposited on the steel substrate. The microstructure, corrosion behavior, and tribological properties of the coating were systematically characterized using scanning electron microscopy (SEM), electrochemical techniques, and tribometry. The results reveal that the coating exhibits a continuous triple-layer structure, consisting of the steel substrate, an intermediate Fe-Al intermetallic compound layer, and an outer aluminum-rich layer. In a 3.5 wt.% NaCl solution, the coating formed a protective Al2O3 film, demonstrating clear passivation behavior. It significantly enhanced the substrate’s performance, achieving an approximately 90% reduction in wear rate and a substantial increase in charge transfer resistance. The coated sample showed a lower friction coefficient (0.24) compared to the bare substrate (0.34). Herein, this work demonstrates that a straightforward and industrially viable hot-dip aluminizing process can effectively improve the corrosion and wear resistance of medium-carbon steel. The findings provide a practical surface-hardening strategy for such steels operating in aggressive environments.

Introduction DNA N6-methyladenine (6mA) is an important epigenetic modification that plays a critical role in gene expression regulation and has been associated with diverse biological processes and diseases. Accurate identification of 6mA sites is essential for understanding its functional significance. Although an increasing number of computational approaches have been proposed, they almost exclusively rely on sequence-derived features. The potential of novel feature representations to further enhance predictive performance remains an important research problem. Methods In this study, we propose FSFT6mA, a novel deep learning-based framework designed to improve 6mA site prediction through feature synthesis. The model is initially trained on the original datasets using a deep convolutional neural network. Subsequently, a Generative Adversarial Network (GAN) is employed to generate synthetic features from intermediate network layers, which are then used to fine-tune the well-trained model in the first stage. Results Incorporating GAN-generated features leads to notable performance gains, improving MCC by 2.6% on A. thaliana and 1.9% on D. melanogaster compared with the base models without synthetic features. Independent validation experiments demonstrate that FSFT6mA achieves superior performance compared to existing state-of-the-art predictors, attaining AUC values of 0.969 and 0.968 on A. thaliana and D. melanogaster , respectively. Discussion These results indicate that FSFT6mA is an accurate tool for DNA 6mA site prediction. The data and the codes used in this study are freely accessible on GitHub ( https://github.com/YuHong-Jin/FSFT6mA ).

Harnessing natural evaporation offers a sustainable pathway for next-generation energy technologies. We present a unified physical and experimental framework for evaporation-driven hydrovoltaic (EDHV) systems that decouples and controls the key interfacial processes underlying electricity generation from heat and sunlight. An intermediate ion-conducting layer separates the evaporative top interface from the silicon-dielectric nanopillar array, enabling independent modulation of evaporation, ion transport, and interfacial chemical equilibrium. This strategy enhances performance and clarifies mechanisms governing thermal and photo-induced charge generation, improving ion migration and electricity output. We develop a predictive equivalent-circuit model that captures process coupling through an analytically derived transfer capacitance. Our results show that capacitive photocharging and thermally modulated surface equilibria-rather than faradaic or photothermal effects-dominate energy conversion. The device achieves 1 V open-circuit voltage and 0.25 W/m² power density, with silicon doping and dielectric choice further boosting performance. These findings inform EDHV optimization across environmental and material conditions.

The benefits and advantages of using 3D gradient printing to create ABS plastic-based layered composites reinforced with carbon fibers (CF) and iron oxide nanoparticles (NP) were studied. Samples with different additive contents, as well as a gradient sample with a gradual change in composition (additive content, wt.%: 30 CF, 15 CF, 0, 5 NP, 15 NP), were 3D printed. The thermophysical, mechanical, and magnetic properties of all samples were determined, and the influence of both qualitative and quantitative composition on them was analyzed. Based on the 3D modeling and the preliminary analysis of interlayer adhesion, the need for an intermediate layer of pure polymer between the layers with CF and NPs was shown. Optimal 3D printing settings were selected, and a part with a gradient composition was manufactured and subsequently used to assemble an industrial robot. Overall, the results reveal that deliberate incorporation of various materials or additives into specific regions of a composite offers a way to realize unique combinations of its mechanical, thermal, and electrical properties, tailored to specific applications.

Purpose: To develop a new domain generalization (DG) framework for diabetic retinopathy (DR) classification using Vision Transformers (VTs). Methods: A deep-learning-based prediction-softening mechanism has been used to allow self-distillation, whereby the knowledge extracted is transferred to intermediate layers. Intermediate representations are further improved with an adaptive convex combination of one-hot labels and internal predictions of the model, which helps regularize the network and counteract overfitting. Various publicly available DR datasets, such as APTOS and EYEPACS, in multi-source and single-source DG settings containing three VT backbones (DeiT, T2T-VT, and CvT) were used. Results: The proposed framework outperforms current DG methods, with better top-1 accuracy, strong calibration and predictive accuracy in unfamiliar domains. Conclusion: The Softening Predictions for Self-Distillation (SPSD)-VT framework is a robust, reliable, and generalizable framework that classifies DR and underscores the important role that domain generalization plays in medical imaging and sets a standard for future studies.

"Sandwich" structured nanorods for timely antibiosis, immunoregulation and neuroangiogenesis to accelerate osteogenesis of Zn-based implants in diabetics.

Electrochemical oxidation degradation of polystyrene nanoplastics by Sm-Mn intermediate layer Ti/Sb-SnO2 anode: Composite metal elements enhance electron transfer and promote the generation of hydroxyl radicals.

Sustainable epoxy composites from hemp/pineapple/glass fibers for lightweight automobile panels.

Role of directly grown multiwalled carbon nanotube as intermediate layers for GaN on c-plane sapphire

Intermediate layer engineering with composite sols for enhanced separation efficiency and hydrothermal stability of 1,2-bis(triethoxysilyl)methane-derived hybrid silica membranes

Remarkably improved electrochemical and safety performances of lithium-sulfur batteries via BaSO4@Ketjen black double-layer-modified separators.

Three-layer trap: Congener-specific PCBs accumulation driven by the biological pump in the Sea of Japan.

Confined polymerization, as an innovative polymerization strategy, achieves precise control over the reaction pathway and microscopic structure of the product by confining the polymerization reaction within the physical space of a micro-nano scale. Compared with traditional large-scale or solution polymerization, confined polymerization is carried out in confined spaces, such as nanochannels, layered intermediate layers, or porous material pores, significantly altering properties such as the polymerization rate, molecular weight distribution, glass transition temperature, and product morphology. This review systematically classifies the limited-domain polymerization strategies in different dimensional spaces, clarifies their mechanism differences, and emphasizes the progress in characterisation techniques, including in situ microscopy, spectroscopy, and computational simulation. Additionally, we discuss confined polymerization in cutting-edge applications, such as water purification, medical diagnosis and treatment, energy storage, catalysis, and composite coatings. By combining fundamental principles with functional innovation, we identify the key challenges, such as real-time mechanism detection and scalable synthesis, and propose future directions, including dynamic limitations, biomimetic design, and AI-driven optimization. The aim of this article is to stimulate the attention of more scholars to the field of confined polymerization, thereby accelerating breakthrough progress in this field and providing innovative material solutions for global challenges such as climate change, disease treatment, and clean energy.