The safety risks associated with urban rail transit equipment are characterized by multi-source heterogeneity and dynamic evolution. Traditional expert-driven static management models often fail to meet the proactive prevention demands in complex scenarios, leading to critical issues such as ambiguous risk identification and insufficiently targeted prevention measures. This study proposes a novel risk assessment and inference method that integrates knowledge graphs with Bayesian networks. First, a safety risk knowledge graph is constructed based on historical accident case reports. Then, a mapping method is proposed to convert the knowledge graph into a Bayesian network. Subsequently, data-driven statistical approaches are employed to estimate the network parameters. Finally, a case study involving equipment failures in urban rail transit is conducted to validate the proposed method. Experimental results demonstrate that the proposed method effectively identifies key risk factors and accurately traces accident causes through backward inference. The method also significantly outperforms traditional approaches in terms of practical accuracy. The findings provide intelligent decision support for the risk management of urban rail transit equipment.
Cross-view geo-localization between drone and satellite images is severely challenged by rapid weather variations, which induce appearance shifts, occlusions, and texture degradation. Inspired by human foveal attention, we propose the Fovea Attention Network (FANet), a robust dual-branch framework comprising: 1) the Weather-Adaptive Global Branch (WAGB) that explicitly injects weather cues (e.g., 'rain/snow') into the feature space via a style-modulation encoder, then captures large-scale structural consistency through a Learnable Region Reassembly (LRR) mechanism; and 2) the Local Semantic Attention Branch (LSAB) that leverages a pretrained segmentation model to generate high-confidence masks, distilling discriminative features from salient regions. An adaptive fusion strategy module fuses global context with fine-grained semantic cues. We further adopt multi-weather adaptive training, treating weather types as related tasks with shared parameters to reduce cross-weather confounding. Extensive experiments on University-1652, SUES-200, and CVUSA show that FANet achieves competitive Recall@1 across all conditions, attaining the highest overall mean with the lowest variance. Notably, it improves Recall@1 by 6.79% under severe low-illumination ('dark') conditions, demonstrating robustness and stability. Our code is available at https://github.com/Jahawn-Wen/FANet.
Somatic mutations accumulate throughout life and have been hypothesized to drive organismal decline. Yet whether these mutations are distributed randomly or whether cells shield their most critical components has remained unresolved. Here we analyze over a million somatic mutations across thirteen human tis-sues, finding that the aging genome exhibits organized vulnerability, captured by the existence of hypo-mutated genes and longevity-associated pathways that have significantly lower mutation burden. Highly connected network hubs are systematically protected from mutation, while peripheral, condition-specific genes accumulate disproportionate burdens. We show that this organized vulnerability arises from the interplay of two independent mechanisms: transcription-coupled repair, and selective filtering. Finally, we validate our findings under experimental mutagenesis, demonstrating intrinsic mechanisms of protection rather than tissue-specific confounders. These findings reframe the somatic mutation hypothesis: organismal decline may not reflect total mutational burden, but where those mutations fall within the cellular network.
Employee knowledge hiding severely hinders organizational innovation, yet how a positive organizational care climate inhibits this behavior remains insufficiently understood. Most existing single-theory explanations fail to clearly articulate the complete psychological pathway linking environmental resources to individual behaviors. This study integrates Conservation of Resources Theory and Affective Events Theory to systematically construct and test a serial mediation model, illuminating how organizational care reduces knowledge hiding via organization-based self-esteem and employee loyalty. A two-wave questionnaire survey was conducted among employees from enterprises in Shanghai, Beijing, Hubei, Henan, and Guangdong, China, with a 6-month interval between data collections. The survey yielded 335 valid responses. Structural equation modeling was used to analyze the data and test the hypothesized model. The results show that organizational care is significantly and negatively related to knowledge hiding behavior. While organization-based self-esteem does not have a significant direct mediating effect, employee loyalty acts as a significant mediator. Furthermore, organization-based self-esteem and employee loyalty play a chain-mediated role in the relationship between organizational care and knowledge hiding. Specifically, organizational care enhances organization-based self-esteem, which in turn increases employee loyalty and ultimately reduces knowledge hiding behavior. This study demonstrates that organizational care effectively reduces knowledge hiding behavior by enhancing employees’ organization-based self-esteem and strengthening their loyalty to the organization. These findings enrich the theoretical frameworks of resource conservation and affective event theories and provide practical guidance for organizations to manage negative knowledge behaviors by fostering a supportive organizational environment.
Boron neutron capture therapy (BNCT) is a binary cancer treatment strategy that relies on the selective delivery of a 10B-containing drug followed by thermal neutron irradiation, generating localized high-LET particles that destroy tumor cells. L-BPA is the most recent and currently the only marketed boron carrier; however, its poor tumor retention and low solubility significantly limit therapeutic efficacy. Thus, developing improved L-BPA derivatives, as well as tools for efficient monitoring of L-BPA biodistribution, is critically important. Herein, we report a turn-on fluorescent probe, WF324, which undergoes rapid and selective fluorescence enhancement upon interaction with L-BPA. WF324 exhibits a pronounced fluorescence turn-on response at 578 nm (λex = 488 nm) along with a obvious Stokes shift of 90 nm, the probe displays a rapid response with a half-response time (t₁/₂) of 12 s and reaching a stable plateau within 150 s in aqueous solution and shows excellent selectivity with minimal interference from metal ions. The probe exhibited a good linear response toward L-BPA in the concentration range of 0-6 μM, with a detection limit as low as 80.48 nM. In real samples, the sensor achieved recoveries of 90.40-103.11% with RSDs of 0.78-3.96% for L-BPA detection in human urine. These properties make WF324 a promising tool for sensitive detection and real-time monitoring of L-BPA, offering valuable support for future BNCT research and development.
This study was conducted to investigate the travel behavior of residents in a medium-sized Chinese city, with the goal of exploring travel characteristics and identifying the key factors influencing urban travel mode choices. While traditional discrete choice models are known for their strong interpretability, their predictive accuracy remains limited. In contrast, machine learning models are recognized for offering higher predictive accuracy but are frequently criticized for their lack of interpretability. To address this issue, a CART-Apriori predictive model was constructed through the integration of the Classification and Regression Tree (CART) model and the Apriori algorithm. Accuracy, the Kappa coefficient, and the Macro-F1 score were utilized as performance metrics for the quantitative comparison of the CART-Apriori model with various alternative models. Additionally, the RuleFit model was employed to extract nonlinear relationships generated by the CART-Apriori model. These relationships were subsequently converted into rule-based features and incorporated into a multinomial Logit linear model to identify the most influential travel rules for each travel mode. The results demonstrated that an average overall prediction accuracy of 82.77% was attained by the CART-Apriori model. Using the Apriori association rule algorithm, the most critical factors influencing urban residents' travel mode choices were ranked in descending order of importance as travel distance, travel purpose, car ownership, and the number of transfers. When walking was chosen as the travel mode, travel distance played a dominant role. When shared electric vehicles or private cars were chosen, travelers were primarily motivated by the intention to reach their destinations directly. Shared bicycles were predominantly chosen by commuters traveling 1-3 km. For bus users, travel distance and the number of transfers were the most influential factors. Ride-hailing users were primarily commuters traveling 1-3 km who required multiple transfers on public transportation.
Hollow structured composites with various elemental compositions and complex shell architectures exhibit great potential in application of heterogeneous catalysis. However, the synthesis of bimetallic alloy nanoparticles (NPs) supported by hollow carbon nanocages is a serious challenge. Herein, we report a streamlined synthesis of hollow spherical Ni/Ru bimetallic NPs embedded in N-doped hollow carbon nanocages (Ni/Ru@HNCs) via a carbonized organic polymer-mediated hard-templating strategy. Core-shell structured SiO2@polydopamine-Ni2+ (PDA-Ni2+) composites are firstly prepared through a one-pot method. Subsequent high-temperature carbonization and KOH etching generate Ni@HNCs featuring a hollow architecture, high specific surface area, abundant active sites, and efficient mass transport. Ru species are then introduced hydrothermally, where a spontaneous galvanic replacement reaction between metallic Ni and Ru3+ ions drives the in situ formation and uniform anchoring of Ni-Ru alloy NPs on the N-doped carbon framework. The resultant Ni/Ru@HNCs exhibit exceptional peroxidase-mimicking activity, efficiently catalyzing the oxidation of 3,3',5,5'-tetramethylbenzidine (TMB) to yield a vivid blue product (oxTMB). Capitalizing on this robust catalytic performance, a sensitive colorimetric sensing platform is established for the rapid and selective detection of tea polyphenols (TP). This work not only demonstrates a rational design of high-performance bimetallic nanozymes with precisely controlled structure and composition but also provides a versatile, scalable strategy for developing advanced enzyme-mimetic materials for biosensing and food safety applications.
With the development of large language models (LLMs), numerous studies have demonstrated their vulnerability to carefully crafted jailbreak attacks. However, existing mitigation measures rarely balance model usability with significant protective effects, raising concerns about model abuse. To address this, we introduce CoT Defender. It preemptively occupies the model's first few generated tokens with a chain-of-thought analysis that hinders attackers from steering the output towards harmful content. We designed a two-stage training framework that strengthens security while preserving usability. Stage 1 fine-tunes the model to follow a structured chain-of-thought format before answering. Stage 2 employs reinforcement learning to refine this reasoning. An auxiliary attacker model continuously synthesizes new jailbreak prompts, and a lightweight evaluator-Probabilistic Structured Output Evaluation (PSOE)-supplies fine-grained rewards by scoring both sentence-level intent capture and token-level format fidelity. We conducted a series of experiments on four models and six attack methods. Across all models, we successfully reduced the average attack success rate to below 8.0 %, with no more than a 7.0 % impact on the response rate for benign requests. Code is available here. Warning: This paper contains red-team data that may be offensive!
Engineering organoids that faithfully replicate the intricate architecture and region-specific features of bodily organs and extraembryonic tissues remains a significant scientific challenge. Previously, we demonstrated that craniofacial skin organoids (cSkOs), containing epidermis, dermis, and hair, could be generated by co-developing epidermal progenitors with cranial mesenchyme. Building on this approach, we precisely adjusted cellular composition and signaling environments to generate ventral skin or-ganoids (vSkOs) with lateral plate mesoderm (LPM) progenitors, successfully recapitulating features of abdominal or groin skin. Modulating early BMP and FGF signaling redi-rected these vSkOs toward an extraembryonic fate, producing human amnion-like tis-sues, termed Amnioids. Like native human amnion, Amnioids rapidly expanded into large, avascular, hairless cysts, in sharp contrast to the primitive vasculature and abundant hair follicles of vSkOs. Single-cell RNA sequencing identified divergent molecular signatures and developmental trajectories, highlighting key roles for NOTCH, WNT, and YAP/Hippo signaling pathways. Functional studies further underscored mesenchymal-epithelial interactions and mechanical forces as critical regulators of epithelial expansion. Together, these models provide potent tools to investigate human development at the embryonic-extraembryonic interface, offering critical insights into congenital skin and amniotic disorders and opening new avenues for precision regenerative therapies.
Cardiac fibrosis represents the terminal pathological progression of diverse cardiovascular diseases, characterized by aberrant activation and migration of cardiac fibroblasts, as well as excessive and disordered deposition of the extracellular matrix. Our previous study showed that NAP1L1 is an important regulator of cardiac fibrosis and is upregulated in ischemic cardiomyopathy patient hearts. Accordingly, discovery of NAP1L1 inhibitors and elucidating their underlying mechanisms of anti-cardiac fibrosis should be urgently needed. Herein, we identified a new NAP1L1 small molecule inhibitor Z1149421873 (named Z11) by the structure-based drug design strategy. Z11 was shown to inhibit cardiac fibroblasts activation, deposition of collagen hypersecretion, and alleviate cardiac fibrosis in both in vitro models induced by TGF-β1 and in vivo myocardial infarction (MI) mouse models. Mechanistically, Z11 interfered with the interaction between NAP1L1 and YAP1, which in turn promotes the ubiquitination degradation of YAP1, thereby inhibiting the AKT/mTOR signaling pathway, and attenuating myofibroblast activity and cardiac fibrosis, and improving cardiac function after MI. This finding may provide new insights for the development of promising candidate drugs for the treatment of cardiac fibrosis related diseases in the future.
Modulating intratumoral copper homeostasis to trigger cuproptosis is a promising copper‑based anticancer strategy, but its clinical translation is hindered by the absence of precise spatiotemporal control over intracellular copper levels and ATP7A‑mediated active copper efflux. Here we report an ultrasound-activated copper molybdate nanoregulator, HCu1.5MoO5 (HCMO), that releases copper on demand and reprograms copper flux in tumor cells under ultrasound. Once internalized, ultrasound activates HCMO to release copper ions, providing a sustained intracellular copper supply. In parallel, HCMO also generates reactive oxygen species that impair mitochondrial function and depress cellular ATP, thereby attenuating the activity of the ATP-dependent exporter ATP7A and narrowing copper efflux. This dual-axis imbalance, in situ copper supplement, and efflux limitation significantly disrupt cellular copper homeostasis, interrupt the tricarboxylic-acid cycle, induce mitochondrial dysfunction, further enhance copper accumulation, and eventually cause irreversible cuproptosis. Together with apoptosis and necroptosis, the coordinated damage releases damage-associated molecular patterns and induces immunogenic cell death. Overall, this spatiotemporally programmable strategy links materials, physical fields, and immunity in a closed framework to coordinately disrupt copper homeostasis, offering a generalizable route to on-demand cuproptosis therapy.
Recovering butyrate from organic waste enables its high-value conversion, aligning with the principles of a circular economy. Traditional butyrate fermentation emphasizes carbohydrates and protein degradation, with limited focus on chain elongation (CE). This study, for the first time, systematically evaluated the effects of different Saccharomyces cerevisiae (SC) concentrations (1, 2, 4, 6, and 8 g/L) on ethanol production (a key electron donor) and subsequent CE for butyrate synthesis, identifying 2 g/L as the optimal SC dosage. At this concentration, butyrate production reached 15.41 ± 2.84 g COD/L, which was 2.72 times higher than that of the blank. Metabolic pathway analysis revealed that yeast not only enhanced substrate degradation (＞90 %) but also facilitated the in situ generation and utilization of ethanol. 16S rRNA indicated 54.10 % relative abundance of butyrate-producing bacteria (Clostridium). Long-term tests found that adding SC reversed the halt in production from prolonged distiller yeast inoculum, stabilising output at 15 g COD/L. Metagenomic analysis revealed that SC inoculation primarily enriched Clostridium luticellarii and Clostridium tyrobutyricum. In addition to raising reverse β-oxidation gene abundance, this treatment also enhanced lactate utilization genes, thereby strengthening acetyl-CoA to butyrate conversion. Through further experiments involving different electron donor ratios and long-term operation, this study highlights the critical role of yeast-bacteria synergy in enhancing butyrate synthesis, providing a theoretical foundation and technical strategy for food waste valorization in line with circular economy principles.
The architectural optimization of multifunctional components within MoS2-based nanohybrids demonstrates potential for amplifying cooperative catalytic effects in biomimetic catalytic systems. Employing APTES@MoS2 composites as specialized sorbents, their inherent amine coordination ability and surface negative charge were exploited to investigate their efficacy in adsorbing transition metal ions such as Ni2+, Fe3+, and Co2+ ions. High-temperature carbonization treatment gradually converted the amorphous APTES polymer into N-doped carbon (denoted as CAPTES), while simultaneously enhancing crystallinity and enabling transition metal doping in MoS2 nanosheets (NSs) through thermal treatment, leading to enhanced catalytic performance. Dense decoration of palladium nanoparticles (Pd NPs) is incorporated to amplify the hydrophilicity and the catalytic efficiency of CAPTES@Fe-MoS2 nanocomposites. Benefiting from the hierarchical hollow configuration with extensively exposed edge sites, optimized charge transfer dynamics, and homogeneously distributed active centers, the engineered tubular heterostructures demonstrate exceptional catalytic performance in both 4-nitrophenol (4-NP) reduction and biomimetic enzymatic reactions. This template-directed synthesis strategy provides a scalable platform for manufacturing heterostructured MoS2-based materials with tunable synergistic functionalities, enabling advanced catalytic solutions for environmental and analytical applications.
Cross-view geo-localization (CVGL) presents significant challenges due to the drastic variations in perspective and scene layout between unmanned aerial vehicle (UAV) and satellite images. Existing methods have made certain advancements in extracting local features from images. However, they exhibit limitations in modeling the interactions among local features and fall short in aligning cross-view representations accurately. To address these issues, we propose a Multi-Scale Cascade and Feature Adaptive Alignment (MCFA) network, which consists of a Multi-Scale Cascade Module (MSCM) and a Feature Adaptive Alignment Module (FAAM). The MSCM captures the features of the target's adjacent regions and enhances the model's robustness by learning key region information through association and fusion. The FAAM, with its dynamically weighted feature alignment module, adaptively adjusts feature differences across different viewpoints, achieving feature alignment between drone and satellite images. Our method achieves state-of-the-art (SOTA) performance on two public datasets, University-1652 and SUES-200. In generalization experiments, our model outperforms existing SOTA methods, with an average improvement of 1.52% in R@1 and 2.09% in AP, demonstrating its effectiveness and strong generalization in cross-view geo-localization tasks.
Recovery of medium-chain carboxylic acids (MCCA) from food waste is constrained by low efficiency and instability. This study validated a short-term aerobic pretreatment (AP) strategy to enhance fungi-bacteria synergy. In batch tests, AP (0.2 vvm) achieved optimal caproate titers of 22.32 ± 1.56 g COD/L. The pretreatment enriched ethanol-producing yeasts and lactate-producing bacteria, establishing a robust co-electron donor pool. Metagenomic analysis revealed that this synergy suppressed the competing tricarboxylic acid cycle, redirecting carbon flux towards reverse β-oxidation (RBO) pathway and providing essential precursors for Clostridium_sensu_stricto_12. In a 134-day semi-continuous operation, AP sustained high titers (17.2-22.1 g COD/L) through a specialized guild dominated by the Ruminococcaceae bacterium BL-6, avoiding the systemic performance deterioration observed in controls. Life cycle assessment (LCA) confirmed a >60% carbon footprint reduction compared to conventional routes. Short-term aerobic pretreatment effectively regulates microbial succession to stabilize low-carbon MCCA production from food waste.
The advancement of highly efficient, selective catalysts stands as a pivotal challenge in methanol steam reforming (MSR) for hydrogen production. In conventional Ni-based catalysts, metallic Ni0 acts as the active site, synergizing with abundant surface oxygen vacancies of the oxide support and generating the intermediate *OCH₂ tends to directly decompose into CO, which further reacts with *H generated from H₂O activation to form CH₄, resulting in poor product selectivity. Herein, we propose a novel Ni/In₂O₃ catalyst that significantly enhances the selectivity of CO₂, increasing it from less than 10% to 90%. Comprehensive characterizations including XRD, XPS, H₂-TPR, UV-vis and TPSR reveal that, unlike conventional Ni0-based catalysts, the Ni/In₂O₃ catalyst forms a very active and selective Ni₂In₃ alloy phase that enables highly dispersed Ni species while endowing the catalyst with excellent capability for H₂O activation. Combined with further theoretical calculations, the catalytic reaction mechanisms of the CH₃OH and CH₃OH-H₂O systems on the catalyst surface are elucidated in detail. Density functional theory (DFT) calculations demonstrate that the d-p orbital hybridization in the Ni₂In₃ alloy effectively alters the electronic structure of Ni species and tunes the adsorption strength of reaction intermediates and balances the surface coverage of CH₂O* and *OH species, facilitating the critical HO-CH₂O* coupling step for CO₂ generation during methanol steam reforming and thereby improving the CO₂ selectivity of the Ni/In₂O₃ catalyst. The study presented in this work indicates that the formation of Ni-In active alloy phase may effectively overcome the inherent product selectivity issue of conventional Ni-based catalysts by regulating the surface coverage of the critical intermediate and therefore reaction pathway during methanol steam reforming.
Controlling catalyst microenvironments using proton shuttles and hydrogen bond donors in the secondary coordination sphere is a promising approach for developing catalysts that can affect multiproton and multielectron transfer processes. In this context, three palladium calixpyrrole complexes with pendent amine (1), amide (2), and carbamate (3) groups were examined as electrocatalysts for the hydrogen evolution reaction (HER). Building on prior studies showing that the palladium complexes generated catalytically active heterogeneous HER catalysts in the presence of p-toluenesulfonic acid monohydrate, 1-3 were evaluated using the significantly milder proton source anilinium tetrafluoroborate. The active catalytic species for all three systems was found to be solution-based, and kinetic analysis uncovered a first-order dependence on acid, as well as large H/D kinetic isotope effect values, which were consistent with proton-coupled electron transfer being rate-limiting. The calixpyrrole complexes displayed exceptional activity, achieving kobs and turnover frequency values of 4.65 × 106 s-1, 4.19 × 106 s-1, and 3.09 × 106 s-1 for 1, 2, and 3, respectively. These catalytic activities and rate constants approached the diffusion rate limit and ranked among the fastest HER catalysts to date.
The evolution of herbivory is one of the most important ecological events in the evolution of terrestrial vertebrates and impacted the ecosystems they inhabited. Herbivory independently developed in a number of tetrapod clades during the Late Carboniferous and Permian, eventually leading to the establishment of the basic structure of modern terrestrial ecosystems. Here we describe a Late Carboniferous pantylid 'microsaur', Tyrannoroter heberti gen. et sp. nov., with expansive occluding palatal and coronoid dental batteries. The shape of the teeth, as revealed by high-resolution micro-computed tomography data, indicates wear from both shearing and grinding motions consistent with herbivory. New data from historical pantylid fossils show that similar adaptations can be traced back as far as the Bashkirian (~318 million years ago), indicating that terrestrial herbivory was already widespread within this group, and originated rapidly following the terrestrialization of tetrapods. The placement of recumbirostran 'microsaurs' on the amniote stem suggests that terrestrial herbivory is not an amniote innovation, although the phylogenetic position of 'microsaurian' tetrapods remains uncertain. Under any phylogenetic scenario, the data presented here reveal that pantylids acquired adaptations to herbivory independently, probably via durophagous omnivory, feeding on insects, shelled animals and tough plant material.
Point-of-care testing (POCT) is crucial for disease diagnosis, especially in resource-limited settings with inadequate laboratory infrastructure. Although visual distance-based readout sensors present a promising approach for instrument-free quantification, their widespread application is hindered by complex fabrication processes and dependence on external driving mechanisms. Here we developed a universal, instrument-free visual quantification platform based on a gold nanobipyramid -loaded agarose (AuNBP/agarose) hydrogel encapsulated in a polycarbonate (PC) tube. The iodine-mediated etching of AuNBPs transduced target recognition into a measurable color-changing hydrogel height. The addition of acid phosphatase (ACP) initiated the enzymatic production of ascorbic acid (AA), driving the etching process and generated a height signal proportional to ACP concentration in the range of 0.1-1000 mU/mL, with a detection limit of 9.72 μU/mL. Moreover, it exhibited high selectivity and satisfactory recoveries for ACP in human serum samples. For DNA detection, target hybridization introduced horseradish peroxidase (HRP), which effectively inhibited etching by consuming H2O2. This strategy enabled sensitive detection of the BRCA1 gene within a concentration range of 100-1500 nM, with a detection limit of 270 nM. Both assays enabled quantitative readout using a simple ruler, without the need for complex instrumentation, demonstrating a universal and cost-effective strategy for POCT detection.