Abstract: Large Language Model (LLM) agents are increasingly deployed to interact with untrusted external data, exposing them to Indirect Prompt Injection (IPI) attacks. While current black-box defenses (i.e., model-agnostic methods) such as “Sandwich Defense” and “Spotlighting” provide baseline protection, they remain brittle against adaptive attacks like Actor-Critic (where injections evolve to better evade LLM’s internal defense). In this paper, we introduce Real User Instruction (RUI), a lightweight, black-box middleware that enforces strict instruction-data separation without model fine-tuning. RUI operates on three novel mechanisms: (1) a Privileged Channel that encapsulates user instructions within a cryptographic-style schema; (2) Explicit Adversarial Identification, a cognitive forcing strategy that compels the model to detect and list potential injections before response generation; and (3) Dynamic Key Rotation, a moving target defense that re-encrypts the conversation state at every turn, rendering historical injection attempts obsolete. We evaluate RUI against a suite of adaptive attacks, including Context-Aware Injection, Token Obfuscation, and Delimitation Spoofing. Our experiments demonstrate that RUI reduces the Attack Success Rate (ASR) from 100% (undefended baseline) to less than 8.1% against cutting-edge adaptive attacks, while maintaining a Benign Performance Preservation (BPP) rate of over 88.8%. These findings suggest that RUI is an effective and practical solution for securing agentic workflows against sophisticated, context-aware adversaries.

Abstract: Abstract To assess the electromagnetic effect on the children’s sensitive organs when they using the smartwatches, this study analyzes the electromagnetic dose absorbed by the children, especially calculate the combind electromagnetic exposure, We propose a multi-frequency smartwatch antenna (GPS L1, LTE FDD band1, 2.4G Wi-Fi/Bluetooth) and construct a 6-year-old human model including ten tissues from three sensitive organs (brain, eyes, heart). The Specific Absorption Rate (SAR) distributions are simulated under speaking and listening postures at different frequencies using CST Studio Suite. Furthermore, the SAR distributions under multi-frequency combined electromagnetic exposure are evaluated to investigate the superposition effects of electromagnetic fields. The results show that, when the child uses the smartwatch for 4G communication in the listening posture, the maximum local SAR reaches 1.683 W/kg, which is nearly half of the ICNIRP exposure limit. Electromagnetic radiation in the child’s brain is mainly concentrated on the surface of the cerebrum and cerebellum, while the radiation in the eyes is mainly distributed in the outer and anterior tissues. Under combined electromagnetic exposure, the SAR values of tissues increase by 1.17 to 3.53 times compared with single-frequency conditions but remain within safety limits. Considering the long-term use of smartwatches, the potential health risks to developing children still deserve attention.

Abstract: Astaxanthin (AST), despite its high bioactivity, exhibits poor stability and low bioavailability due to its strong lipophilicity and inherent degradation susceptibility. To overcome such a challenge, we developed a food-grade oleogel delivery system using a soy protein-arabinoxylan (SA) glycosylated complex modulated by different concentrations (0.5%–3%) of sucrose ester (SE) or soy lecithin. We show that the emulsifier concentration has a non-linear effect on the oleogel microstructure: an optimal level of 1% had a significant impact on the interfacial compactness and network density, giving rise to improved thermal stability, rheological strength and AST encapsulation efficiency (81.27%). During in vitro digestion, the SA matrix in combination with emulsifiers allowed gastric protection and intestinal-targeted release of AST with a bioaccessibility of up to 88.84% (SAO-SE-AST). This controlled-release profile directly translated into enhanced in vivo antioxidant efficacy in wild-type Bristol N2 Caenorhabditis elegans as evidenced by reduced lipofuscin accumulation, elevated thermotolerance (survival rate: 64.44%-73.33%), suppressed reactive oxygen species levels and activation of endogenous antioxidant enzymes (superoxide dismutase as well as glutathione peroxidase). Collectively, this research has uncovered that food-grade emulsifiers are not only stabilizers, but also key regulators of oleogel architecture and bioactive functionality. These results provide a structure-digestion-bioactivity correlation for protein-polysaccharide oleogels, representing a rational design strategy for high-performance delivery systems of lipid-soluble nutraceuticals.

Abstract: In art and architecture, synchronicity—the parallel emergence of related ideas without direct causal links—has long shaped the evolution of design cultures. Recently, heritage architecture across Europe has become an active stage for fashion shows, art exhibitions, and live performances, encouraging audiences to experience buildings not merely as backdrops but as performative, meaning-generating spaces. Italy offers a significant field of observation through projects positioned beside restoration. While not constituting restoration activities in a conventional sense, these interventions engage critically with existing heritage by activating processes of reuse, management, and cultural valorisation. Such strategies contribute to sustainability on multiple levels: they address architectural and conservation constraints while simultaneously supporting environmental preservation, limiting land consumption and urban sprawl, reducing waste, and broadening public access to heritage. Through the analysis of selected case studies, this paper investigates the dialogue between performance art and heritage architecture as a sustainable practice. The discussion is enriched by interviews with three professionals whose work in central Italy exemplifies innovative approaches to engaging communities with their surrounding heritage. The study argues that performative practices can operate as complementary tools to restoration, fostering new forms of cultural sustainability and expanding contemporary understandings of heritage stewardship.

Abstract: Aim: To evaluate anxiety, depression, perceived stress, and sleep quality in the parents of children who underwent liver transplantation in Turkey. The impact of other factors on mental health status were also examined. Method: This was a single-center, cross-sectional study including the parents of 50 children after liver transplantation. Major sociodemographic variables included parental age, sex, education, chronic disease, and immigration status. We also recorded children’s demographics, transplant-related data, follow-up findings, and mental health status. Instruments for psychiatric assess-ment included the Generalized Anxiety Disorder 7-item scale (GAD-7; anxiety), Patient Health Questionnaire-9 (PHQ-9; depression), Perceived Stress Scale-10 (PSS-10; stress), and Pittsburgh Sleep Quality Index (PSQI; sleep quality). Results: We enrolled 50 parents of 50 pediatric liver transplant recipients (28 fathers, 22 mothers, mean age: 40.10 ± 6.65). Time since transplantation showed weak negative correlations with PHQ-9 and GAD-7. Stress (PSS) levels had weak to strong positive correlations with PSQI, PHQ-9, and GAD-7. Sleep quality (PSQI) was positively correlated with PHQ-9 and GAD-7. Depressive findings (PHQ-9) were strongly and positively correlated with GAD-7. High PHQ-9 scores were found to be independently associated with shorter time since transplant (p=0.006) and high PSS (p=0.011). High GAD-7 scores were independently associated with shorter time since transplant (p=0.034) and high PSS (p=0.005). Conclusion: The parents of pediatric liver transplant recipients experience high levels of stress, sleep issues, depression, and anxiety, which demonstrate multiple correlations.

Abstract: Background/Objectives: Rapid Eye Movement (REM) Sleep Behavior Disorder (RBD) is characterized by dream enactment due to reduced physiological muscle atonia during REM sleep and is clinically relevant as a potential prodromal marker for neurodegenerative disorders. This study aims to evaluate whether normalized beta-band power extracted from poly-somnographic signals can differentiate RBD subjects from healthy controls, and to compare the discriminative behavior of C4–A1 EEG versus EMG1–EMG2 channels during REM sleep. Methods: Polysomnographic recordings were obtained from the PhysioNet CAP Sleep Data-base. One-minute epochs were analyzed across sleep stages, with emphasis on REM. Signals were preprocessed to remove DC offset and were windowed with overlap prior to spectral estimation. Short time–frequency analysis of power spectral density (PSD) was applied to compute band-limited power in standard EEG frequency ranges (delta, theta, alpha, beta). Band power values were normalized by total spectral power to derive nor-malized indices. Comparative feature analysis was performed for C4–A1 and EMG1–EMG2 channels. Results: Normalized beta-band power during REM sleep showed clear separation between healthy subjects and RBD patients. In the C4–A1 channel, normalized beta power was higher in RBD than controls (controls: 0.0010–0.0049; RBD: 0.0076–0.014). In the EMG1–EMG2 channel, the difference was more pronounced (controls: 0.0020–0.0089; RBD: 0.053–0.0791). Conclusions: Normalized beta-band power, particularly during REM sleep, is a promising, low-complexity marker for RBD detection. The stronger separation in EMG1–EMG2 sug-gests that targeted channel selection may enhance practical screening pipelines for sleep disorder assessment.

Abstract: Diabetic peripheral neuropathy (DPN) remains a leading cause of disability in diabetes, yet current care is largely symptomatic. Increasing evidence places early dysfunction of the blood-nerve barrier (BNB)—a core element of the peripheral nerve neurovascular unit (PNVU)—at the intersection of metabolic stress and neuroinflammation. This review synthesizes a redox-centered model of BNB failure in DPN: (i) chronic hyperglycemia and dyslipidemia overwhelm endogenous antioxidant defenses, driving reactive oxygen species (ROS) imbalance; (ii) ROS-associated endothelial activation promotes endothelial-immune crosstalk, leukocyte recruitment, and macrophage polarization; and (iii) progressive loss of tight-junction and barrier homeostasis increases paracellular permeability and exposure of nerves to pro-inflammatory and neurotoxic mediators. We then evaluate incretin-based therapies—GLP-1 receptor agonists, DPP-4 inhibitors, and emerging multi-agonists—as candidate PNVU/BNB stabilizers. Beyond glucose and weight effects, these agents may dampen oxidative and inflammatory signaling, enhance antioxidant pathways (e.g., Nrf2), and preserve molecular determinants of BNB integrity via indirect metabolic unloading and potentially GLP-1R-dependent vascular-immune actions. By reframing DPN as a neurovascular-immune disorder driven by redox imbalance, we highlight barrier-focused biomarkers and therapeutic opportunities for disease modification.

Abstract: In large scale image retrieval and big data analytics it is a big challenge to search similar images from high dimensional data. Mostly used algorithms are Locality Sensitive Hashing and Random Projection Based Hashing. They are widely used for approximate nearest neighbor searching. These two algorithms treat all input features uniformly while they ignore feature importance and class separability. In this research we aim to propose a lightweight hashing framework named Adaptive Feature Aware Hashing which integrates feature weighting prior to projection-based hashing. The algorithm computes data-driven feature weights using variance, between-class separability, and Fisher-style discriminative criteria to enhance discriminative power during hash code generation. We also incorporated multi table and multi probe hashing which enhances discriminative power during hash code generation. For this research we used MNISH dataset for experimental evaluation. We compared the results against a Baseline Locality-Sensitive Hashing (LSH) method using random projections. Our results indicate that The AFAH methods (v1 and v2 Fisher) significantly improved both precision and recall compared to the Baseline LSH, with AFAH v2 Fisher showing the highest precision (0.7557) and AFAH v1 having the highest recall (0.2285).

Abstract:

The ‘Tenuifoliae irises’ are a distinctive group of beardless, rhizomatous perennial irises, which are characterised by their somewhat vertical rhizomes, typically clothed at the apex with long maroon-brown, sharp fibrous remains of leaf sheaths; perianth tube long, filiform to scapiform; stigma bilobed; capsules often trigonous to six-ribbed, apically beaked; and seeds angulose to subcubic or pyriform, lacking fleshy appendages, and with testa hard, irregularly wrinkled. The representatives of the aggregate are mostly native to the dry steppes and grasslands from lowland to high mountain habitats of Central and Eastern Asia, extending westwards to the Black Sea and Caspian regions. Morphological classification of the ‘Tenuifoliae irises’ recognises about ten to eleven species, which are arranged into two genera, Sclerosiphon to Cryptobasis. Diverse molecular research recovered members of the ‘Tenuifoliae irises’ in contrasting placements within the ‘Iris-flower clade’. Sometimes, Sclerosiphon was sister to Eremiris, but Cryptobasis aligned with the ‘Spuria irises’ (Chamaeiris) and the ‘Spanish irises’ (Xiphion and related genera); in other cases, both Sclerosiphon and Cryptobasis formed a clade sister to Chamaeiris, or Cryptobasis alone was identified as the basal member of the Iris s.l. clade, positioned immediately after Siphonostylis. To examine these taxonomic discrepancies within a rigorous molecular‑systematic framework and using 12 reliably authenticated specimens, we generated 24 sequences of the matK gene (12) and the trnL (UAA)–trnF (GAA) loci (12) from members of the ‘Tenuifoliae irises’. These sequences were subsequently incorporated into a comprehensive dataset of the ‘Iris‑flower clade’, enabling a broader analytical assessment. The obtained three-taxon statement hierarchy of patterns and maximum likelihood phylogenetic trees both recover the ‘Tenuifoliae irises’ as monophyletic and sister to Chamaeiris, and in turn to the ‘Xiphion s.l. clade’. We also found Sclerosiphon and Cryptobasis as sister genera. The morphological and karyological data supporting those relationships are discussed, which allow getting back to Rodionenko’s sources and recovering Sclerosiphon in his original sense, alongside Cryptobasis. Furthermore, the molecular results allow us expanding Sclerosiphon to include the Eastern Chinese members of the aggregate. In consequence, five new combinations (one series and four species) are established in the genus, one lectotype is designated, and data on nomenclature, distribution and ecology of the accepted species are reported.

Abstract: Cabbage is considered a healthy vegetable due to its chemical composition and high nutritional value. This is due to the presence of carbohydrates and dietary fiber as the main constituents, as well as the presence of vitamin C. The end product thus obtained (sauerkraut) is a low-calorie product with a long shelf-life. The most important role in the fermentation of cabbage is played by lactic acid bacteria whose activity is influenced by physical factors such as temperature and some chemical factors such as salt concentration or the addition of spices which, in addition to their flavoring effect, may also have an inhibitory effect on undesirable microflora. The present study investigates the effect of essential oils extracted from plants on lactic acid bacteria responsible for the fermentation of cabbage. Essential oils from thyme, dill, wild thyme, bay and basil were tested. The obtained results have shown that the essential oils that were added to the fermentation mass in concentrations of 0.015% did not inhibit the activity of lactic acid bacteria responsible for lactic fermentation.

Abstract: This article introduces a tutorial-style implementation of Quantum Link Prediction (QLP) for citation network analysis in historiographical research, with a specific focus on the transnational historiography of Mongol military campaigns. Using a manually curated citation network of Russian and American military treatises from 1875 to 2012, the study applies simulated quantum random walks to identify previously unknown citation pathways. The article is structured to guide researchers through each phase of the QLP workflow, from network preprocessing and quantum circuit construction to result interpretation, making it accessible to scholars in the humanities new to quantum methods. Through this approach, we discover a previously unknown transmission link connecting the Russian and American corpora. This finding not only reshapes the existing citation network but also demonstrates the potential of QLP as an introductory use case for teaching quantum computing to learners in the humanities. To support reproducibility and future adoption the open-source QuantumRandomWalks package was published in conjunction with this paper.

Abstract: Background Generative AI (genAI) has the capacity to create realistic and convincing animal videos, however, it must simplify and reduce behavioural variation to do so, possibly leading to misinformation. Methods We categorised 29 videos in the press release for a specific video genAI engine. Twelve featured animals. We mapped each video to the Five Domains and categorised behaviour and welfare within. Results Negative welfare was rarely seen, ranging from 8% (n = 1) for Nutrition, to 42% (n =5) for Behavioural Interactions. By contrast, Mental State, Environment, and Behavioural Interactions appeared positive in >42% (n = 5) of the videos featured. However, videos were often misleading or did not represent accurate animal behaviour. Limitations This work was limited to a press-release of data and does not explore user experience. Conclusions GenAI videos pose a new route for client confusion and veterinarians need to incorporate genAI misinformation combatting in their practice.

Abstract: Humor and media effects research have a long history together, but there have been few broad-based reviews of that research. A review of 34 experimental research studies was undertaken. Though two-thirds of the studies were guided by theory, only around 21% included a humor theory. Humor was often operationalized using traditional media content, with the humor itself being disparaging humor or satire. Using some measure of humor as a manipulation check was common. However, few studies assessed positive emotion as a response to the humor content at post-exposure. A majority of studies assessed perceived funniness of humor content. Attitudinal and affective outcomes were used frequently, and moderation was included more frequently than mediation. The discussion highlights avenues for greater integration of these research areas.

Abstract: This study investigates the corrosion behavior of a WC-6Co cemented carbide (94 wt% WC, 6 wt% Co) in acidic (pH 2) and alkaline (pH 13) aqueous environments, with em-phasis on implications for reconditioning processes. Both electrolytes, characterized by their high electrical conductivity, are used in industrial electrochemical stripping of PVD coatings. While acidic electrolytes are already established for stripping coatings from hard metal substrates, the influence of the alkaline electrolytes on substrate integrity remains insufficiently explored, especially considering the implication of reconditioning. Elec-trochemical characterization was performed using potentiodynamic polarization method, followed by surface analysis via SEM, EDX, and laser confocal microscopy. Two distinct corrosion mechanisms were identified, corresponding to the respective pH conditions and consistent with predictions from Pourbaix diagrams. In acidic media, cobalt dissolution occurred alongside strong passivation of tungsten through the formation of WO₃. In contrast, under alkaline conditions, tungsten formed soluble tungstate ions (WO₄²⁻), leading to progressive leaching of WC grains, while cobalt exhibited passivation via a Co(OH)₂ layer, mitigating binder degradation. Within the scope of this work, electrolytes used for electrochemical stripping were examined. The investigation focused on their corrosive impact on uncoated hard-metal substrates under electrochemical stripping conditions, as these become exposed to both the electrolyte and applied potential once the coating is removed. Coating removal itself was not addressed. A key finding is that oxide or hydroxide passivation on cemented carbides does not inherently guarantee protection. Its effectiveness depends strongly on the nature of the formed layer. In the acidic elec-trolyte, pseudo-passivation by formation of WO₃ layer initially inhibits corrosion but leads to significant material loss upon its breakdown. These findings provide valuable guidance for the application of cemented carbides in electrochemical stripping processes used for PVD coating removal.

Abstract: Background: This study aims to psychometrically validate the abbreviated version of the Connor-Davidson Resilience Scale (CD-RISC-10) in Peruvian nurses, evaluating its convergent validity through its association with perceived stress and empathy. Methods: A cross-sectional psychometric study was conducted in 374 Peruvian nurses to evaluate the psychometric properties of CD-RISC-10 through confirmatory factor analysis (CFA). In addition, convergent validity was examined by correlational analysis with Spearman's ρ coefficient with empathy and resilience. Results: The CFA confirmed that the one-dimensional model has a good fit (CFI = 0.978, TLI = 0.971, RMSEA = 0.080, and SRMR = 0.044). Cronbach's alpha of 0.89 and McDonald's omega of 0.81 were obtained. Convergent validity showed significant correlations with perceived stress (ρ = -0.23, p < 0.001) and empathy (ρ = 0.31, p < 0.001). Conclusion: The CD-RISC-10 has excellent psychometric properties in Peruvian nurses. Future studies are needed to evaluate their factorial invariance between clinical specialties and determine cut-off points.

Abstract: In this article we analyze a dynamical system known as the FitzHugh-Nagumo model, which offers many characteristics of nonlinear systems, such as bifurcation, excitability or limit cycle. The dynamics associated with sets of values of the parameters associated with this model, called excitable, oscillatory and bistable, are analyzed. Then adding a perturbing or diffusive term to the system through the Laplacian, it is studied how these dynamics propagate in a one-dimensional or two-dimensional extended medium, through the definition of a cellular automaton with periodic initial conditions.

Abstract: Open-pit mining relies heavily on visual inspection to identify indicators of slope instability such as surface cracks. Early identification of these geotechnical hazards allows for the implementation of safety interventions to protect both workers and assets in the event of slope failures or landslides. While computer vision (CV) approaches offer a promising avenue for autonomous crack detection, their effectiveness remains constrained by the scarcity of labelled geotechnical datasets. Deep learning (DL) models require large amounts of representative training data to generalize to unseen conditions; however, collecting such data from operational mine sites is limited by safety, cost, and data confidentiality constraints. To address this challenge, we propose a hybrid game engine—generative artificial intelligence (AI) framework for large-scale dataset synthesis. Leveraging a parameterized virtual environment developed in Unreal Engine 5 (UE5), the framework captures realistic images of open-pit surface cracks and enriches their visual diversity using StyleGAN2-ADA. The resulting datasets were used to train the YOLOv11 real-time object detection model and evaluated on a real-world dataset of open-pit slope imagery to assess the effectiveness of the proposed framework in improving CV model generalizability under extreme data scarcity. Experimental results demonstrated that models trained on the proposed framework substantially outperformed the UE5 baseline, with average precision (AP) at intersection over union (IoU) thresholds of 0.5 and [0.5:0.95] increasing from 0.403 to 0.922 and 0.223 to 0.722 respectively, accompanied by a reduction in missed detections from 95 to eight for the best-performing configurations. These findings demonstrate the potential of hybrid generative AI frameworks to mitigate data scarcity in CV applications and support the development of scalable automated slope monitoring systems for improved worker safety and operational efficiency in open-pit mining.

Abstract: This study investigates the key determinants of firm profitability in the global automotive sector, examining whether superior returns on assets (ROA) stem from operational efficiency, strategic leverage, or innovation intensity, and highlighting the potential trade-off between efficiency and investment in capital-intensive industries. Analysing a global panel dataset of 192 automotive firms from 38 countries over 2010-2024, a fixed-effects regression model with Driscoll-Kraay standard errors was applied to control for unobserved heterogeneity, heteroskedasticity, and cross-sectional dependence across 11 financial and strategic variables. The findings reveal that firm size and inventory turnover are significant positive drivers of profitability, while research and development (R&amp;D) intensity exerts a strong negative impact. The positive association with the effective tax rate reflects reverse causality, where more profitable firms incur higher tax burdens, rather than a causal effect of taxation on performance. Notably, working capital management, leverage, sales growth, and capital expenditure showed no statistically significant effects after controlling for firm and time effects. Temporal fluctuations, including a marked profitability decline in 2024, underscore the sector’s sensitivity to macroeconomic shocks. This study contributes robust, large-scale empirical evidence on the short-term profitability trade-off associated with R&amp;D intensity in a globally integrated industry, addressing cross-sectional dependence through its methodological approach.

Abstract: Finite-size suppression of the Curie temperature (T_c) in ferroelectric perovskite nanostructures remains an important yet insufficiently resolved problem, with reported scaling exponents varying considerably across experimental and theoretical studies. Although density functional theory provides atomistic insight into size-dependent behaviour, its high computational cost limits systematic exploration across broad size ranges. Conversely, purely empirical fitting approaches often lack physical interpretability and formal uncertainty quantification. In this work, a physics-informed surrogate modelling framework is developed to investigate finite-size scaling in BaTiO₃ and KNbO₃ nanostructures using a structured dataset compiled from the literature. The model is based on thermodynamically motivated scaling behaviour, enabling extraction of physically meaningful size-dependent parameters. Bootstrap resampling is employed to quantify statistical robustness, yielding scaling exponents of 1.59 (95% confidence interval: 1.43–1.72) for BaTiO₃ and 1.40 (95% confidence interval: 1.31–1.52) for KNbO₃. Gaussian Process regression is further integrated to provide uncertainty-aware predictions across the nanoscale domain. In addition to forward prediction, the framework enables inverse estimation of the minimum particle size required to preserve ferroelectric stability at a specified operating temperature. For a threshold of 300 K, the predicted critical sizes are approximately 4.96 nm for BaTiO₃ and 2.89 nm for KNbO₃. Extension to a coupled size–strain formulation produces a two-dimensional stability map, demonstrating tunable interactions between confinement and strain. Overall, the proposed methodology provides a transparent, statistically rigorous, and computationally efficient framework for predictive analysis and rational design of nanoscale ferroelectric materials.

Abstract: While enterprises amass vast quantities of data, much of it remains chaotic and effectively dormant, preventing decision-making based on comprehensive information. Existing neuro-symbolic approaches rely on disjoint pipelines and struggle with error propagation. We introduce the large ontology model (LOM), a unified framework that seamlessly integrates ontology construction, semantic alignment, and logical reasoning into a single end-to-end architecture. LOM employs a construct-align-reason (CAR) pipeline, leveraging its unified architecture across all three stages: it first autonomously constructs a domain-specific ontological universe from raw data, then aligns neural generation with this structural reality using a graph-aware encoder and reinforcement learning, and finally executes deterministic reasoning over the constructed topology, node attributes and relation types. We evaluate LOM on a comprehensive benchmark constructed from diverse real-world enterprise datasets. Experimental results demonstrate that LOM-4B achieves 88.8% accuracy in ontology completion and 94% in complex graph reasoning tasks, significantly outperforming state-of-the-art LLMs. These findings validate that autonomous logical construction is essential for achieving deterministic, enterprise-grade intelligence.