Abstract: Transarterial chemoembolization (TACE) is the standard treatment for patients with intermediate-stage hepatocellular carcinoma (HCC), yet nearly half of treated patients fail to achieve durable benefit, and reliable biomarkers enabling early therapeutic stratification are still lacking. Treatment response is typically assessed by imaging one month after TACE and at three-month intervals, potentially delaying timely access to alternative therapies in non-responding patients. Circulating microRNAs (miRNAs) represent promising biomarkers due to their stability in body fluids and ease of detection. Here, we evaluated circulating miR-22 as an early predictor of TACE response and as a mechanistically relevant therapeutic target. Circulating miR-22 levels were measured by microarray and quantitative RT–PCR in three independent cohorts of early-to-intermediate stage HCC patients undergoing TACE. Circulating miR-22 increased significantly in non-responders as early as 48 h after treatment, and fold changes consistently predicted treatment failure across two independent validation cohorts. Mechanistically, we identified the G2/M checkpoint kinase WEE1 as a direct functional target of miR-22. Modulation of the miR-22/WEE1 axis affected cell-cycle progression, proliferation, apoptosis, and DNA damage response in HCC cell lines and xenograft models. Under hypoxia-mimicking conditions combined with doxorubicin exposure, pharmacological inhibition of WEE1 induced mitotic catastrophe in highly proliferative miR-22–silenced cells. Collectively, these findings identify early post-TACE elevation of circulating miR-22 as a biomarker of non-response and highlight the miR-22/WEE1 axis as a potential target for precision treatment strategies in HCC.

Abstract:

In Cambodia, farmers construct artificial household bat roosts to collect and sell guano as fertilizer. We investigated farming practices and attendant spillover risks using: 1) surveys on guano production; 2) estimating bat population size and species present using carcasses, visual identification, and audio recordings; 3) surveying guano-producing and neighbor households on water, sanitation, and hygiene practices; and 4) testing guano and household food, water and surfaces for coronaviruses by PCR. Bat roosts are constructed using dried palm leaves with coconut tree and/or steel/concrete supports. Roosting areas ranged from 42-327 m², bat abundance varied from 0-11,187, guano production was 5-120 kg/week, guano yields were 0.15-0.4 kg/m2/week, and farmers earned ~100-200 USD/household/month. Higher guano production in peak (normally wet) season was associated with greater bat abundance (p=0.016). The lesser Asiatic yellow house bat (Scotophilus kuhlii) was the only bat species identified. Roosts were <20 m from guano-producing households. Neighbors and households’ hygiene risks included not having handwashing stations and not covering food in storage/while drying. Alphacoronaviruses or Infectious Bronchitis Virus were found in 14.6%, 17.3%, 2.9%, 1.4%, and 0.0% of guano, urine, surface, food, and water samples, respectively. While guano farming offers economic benefits, spillover risks exist. Safe guano collection and storage, handwashing, and food covering in guano-producing communities are necessary to mitigate spillover risks.

Abstract: We pose the Typhoon Engulfment Grand Challenge: determine whether any physically realizable feedback control law can guarantee the safe flight and landing of a commercial aircraft under all extreme, physically admissible typhoon wind fields—or prove that such a guarantee is mathematically impossible. The problem is formulated as a two-player zero-sum differential game between an autopilot (the minimizer, seeking survival) and an adversarial but physics-constrained typhoon (the maximizer, seeking to force the aircraft outside its safe operating envelope). We detail the coupled nonlinear dynamics, define the safe operating set in the airspeed–load-factor plane, and identify four interlocking barriers—high-dimensional Hamilton–Jacobi–Isaacs equations, PDE-constrained adversarial disturbances, hybrid structural failure dynamics, and imperfect observations—that place this problem beyond the reach of any existing mathematical framework. This article serves as a formal statement of the challenge, provides accessible explanations for researchers across disciplines, and charts concrete research directions for communities spanning control theory, aerospace engineering, applied mathematics, machine learning, fluid dynamics, structural mechanics, formal verification, operations research, signal processing, meteorology, and independent researchers worldwide.

Abstract:

The green peach aphid, Myzus persicae (Sulzer), is a globally important agricultural pest whose management is increasingly challenged by widespread insecticide resistance, prompting interest in alternative and sustainable control strategies such as endophytic fungi. This study evaluated the effects of two endophytic fungi, Trichoderma harzianum and Chaetomium cupreum, applied individually or as a 1:1 mixture, on the population ecology of M. persicae feeding on capia-type red pepper (Capsicum annuum L.). Aphid development, survival, and reproduction were assessed using age-stage, two-sex life table analysis. Contrary to expectations, T. harzianum significantly enhanced aphid population growth, resulting in a higher intrinsic rate of increase (r = 0.42 d^-1), finite rate of increase (λ = 1.52 d^-1), and net reproductive rate (R₀ = 87.67 offspring) compared to the control (r = 0.32 d^-1, λ = 1.37 d^-1, R₀ = 42.90 offspring). The mixture treatment also increased population parameters, whereas C. cupreum showed limited effects on aphid life table traits. Population projections indicated that T. harzianum treatment could produce aphid populations approximately 380 times larger than the control after 60 days. These results suggest that T. harzianum may improve host plant quality in ways that indirectly favor M. persicae. The findings highlight the importance of evaluating plant–fungus–herbivore interactions before incorporating endophytic fungi into integrated pest management programs.

Abstract: Scaling generalist GUI agents is hindered by the data scalability bottleneck of expensive human demonstrations and the ``distillation ceiling'' of synthetic teacher supervision. To transcend these limitations, we propose UI-Oceanus, a framework that shifts the learning focus from mimicking high-level trajectories to mastering interaction physics via ground-truth environmental feedback. Through a systematic investigation of self-supervised objectives, we identify that forward dynamics, defined as the generative prediction of future interface states, acts as the primary driver for scalability and significantly outweighs inverse inference. UI-Oceanus leverages this insight by converting low-cost autonomous exploration, which is verified directly by system execution, into high-density generative supervision to construct a robust internal world model. Experimental evaluations across a series of models demonstrate the decisive superiority of our approach: models utilizing Continual Pre-Training (CPT) on synthetic dynamics outperform non-CPT baselines with an average success rate improvement of 7% on offline benchmarks, which amplifies to a 16.8% gain in real-world online navigation. Furthermore, we observe that navigation performance scales with synthetic data volume. These results confirm that grounding agents in forward predictive modeling offers a superior pathway to scalable GUI automation with robust cross-domain adaptability and compositional generalization.

Abstract: The present study examines the effects of ICT education on the competency development of 600 trainee teachers in Sri Lanka's National Institutes of Education (NIE). With respect to ICT tools and their interfaces in relation to digital literacy, teaching, and overall facilitator effectiveness, the author presents a discrete quantitative, cross-sectional ICT study. The sample was made up of 300 teachers educated in ICT and 300 teachers educated in traditional (non-ICT) methods. The study utilized a self-administered questionnaire. The author applied and described ICT, regression, and correlation analyses and integrated the ICT variable with effectiveness and competency development in Sri Lanka. The author observed that ICT educators had a higher self-efficacy and ICT tool usage and demonstrated improvements in digital literacy and pedagogy compared to non-ICT educators. A lack of integration of supplemental education technology to support teaching baseline ICT competencies was noted. A holistic approach to teacher training was advocated based on the integrated train of constructivist learning theory and TPACK framework. This study advances the empirical support of ICT and its contribution to the Sustainable Development Goal (SDG) 4, Quality Education, by demonstrating the increased value of educator competencies. The study offers actionable insights to enhance ICT training within the teacher education curricula in Sri Lanka, with the aim of equipping teachers to tackle the challenges of the digital era.

Abstract: This paper addresses the critical communication barrier experienced by deaf and hearing-impaired individuals in the Arab world through the development of an affordable, video-based Arabic Sign Language (ArSL) recognition system. Designed for broad accessibility, the system eliminates specialized hardware by leveraging standard mobile or laptop cameras. Our methodology employs Mediapipe for real-time extraction of hand, face, and pose landmarks from video streams. These anatomical features are then processed by a hybrid deep learning model integrating Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), specifically Bidirectional Long Short-Term Memory (BiLSTM) layers. The CNN component captures spatial features, such as intricate hand shapes and body movements, within individual frames. Concurrently, BiLSTMs model long-term temporal dependencies and motion trajectories across consecutive frames. This integrated CNN-BiLSTM architecture is critical for generating a comprehensive spatiotemporal representation, enabling accurate differentiation of complex signs where meaning relies on both static gestures and dynamic transitions, thus preventing misclassification that CNN-only or RNN-only models would incur. Rigorously evaluated on the author-created JUST-SL dataset and the publicly available KArSL dataset, the system achieved 96% overall accuracy for JUST-SL and an impressive 99% for KArSL. These results demonstrate the system’s superior accuracy compared to previous research, particularly for recognizing full Arabic words, thereby significantly enhancing communication accessibility for the deaf and hearing-impaired community.

Abstract: Brucellosis is an important infectious disease affecting livestock worldwide, causing reproductive losses in small ruminant production systems and representing a persistent challenge for animal health and public health under a One Health perspective. Despite its importance, epidemiological information on ovine brucellosis in Ecuador remains scarce. This cross-sectional study aimed to estimate the seroprevalence of Brucella spp. infection in sheep at both the individual and herd levels and to identify potential risk factors associated with infection. This study represents the largest epidemiological investigation conducted to date on ovine brucellosis in Ecuador, covering the main sheep-producing regions of the country (more than 95% of the national sheep population). Between 2024 and 2025, a total of 970 sheep from 385 farms were sampled. Serum samples were analyzed using a commercial ELISA assay, and epidemiological information was obtained through structured farm surveys. Statistical analyses included bivariate tests and a multivariable model using Generalized Estimating Equations (GEE). The overall individual seroprevalence was 5.1% (49/970), while herd-level prevalence reached 7.5% (29/385). Multivariable analysis identified abortion rate (OR = 1.045; 95% CI: 1.016–1.074) and isolation of sick animals (OR = 2.843; 95% CI: 1.150–7.027) as factors associated with seropositivity, whereas access to technical advisory services was identified as a protective factor. These findings provide essential epidemiological evidence to support surveillance programs and improve biosecurity and veterinary extension strategies in Ecuadorian sheep production systems.

Abstract: Background:Transcrestal maxillary sinus floor elevation (TSFE) is a key technique for posterior maxillary bone deficiency in implantology, with sinus membrane perforation as the main complication. Accurate risk assessment is critical for improving surgical success and implant survival rates. Findings:TSFE surgical risk is determined by six core factors: bone defect type, elevation height, sinus floor morphology, sinus membrane status, ostium patency, and immediate implant placement. Surgical difficulty can be stratified into Easy/Moderate/Difficult, with matching technical strategies to reduce risks. Conclusions: and RelevanceMultidimensional risk stratification and targeted technique selection effectively control TSFE risks. Surgeons with lateral window technique experience are recommended for TSFE, and multidisciplinary collaboration is advised for high-difficulty cases, providing precise guidance for clinical implant practice.

Abstract: This study presents a distribution-optimized mesostructure estimation method for modeling near-surface aggregate size distributions in concrete by optimizing the spatial arrangement of polydisperse spherical aggregates with respect to formwork boundaries. The approach is based on minimizing the deviation between a generated cumulative aggregate volume function and an idealized linear target function corresponding to a constant area fraction along the specimen depth. To enable efficient computation for systems containing a large number of aggregates, grain size classes derived from the grading curve are represented using symmetric Beta distributions, allowing each group to be described by a single shape parameter. The resulting optimization problem is solved using a derivative-free Powell algorithm. The method inherently captures wall effects, leading to a migration of smaller aggregates toward the specimen boundaries to compensate for geometric constraints of bigger aggregates. Experimental validation was performed by determining the depth-dependent mean bulk density of a concrete cube using incremental surface grinding combined with high-resolution 3D laser scanning. The optimized mesostructure shows strong agreement with measured density profiles, significantly improving over a non-optimized distribution. Furthermore, increasing aggregate volume fractions intensify near-surface accumulation of fine particles. The proposed method provides a computationally efficient framework for incorporating wall effects into mesoscale concrete models.

Abstract:

Background/Objectives: Trigeminal neuralgia (TN) is a debilitating neurological condition characterized by recurrent, severe pain linked to peripheral and central sensitization within trigeminal pathways. Although current pharmacologic treatments are limited by inadequate efficacy or dose-limiting side effects, botulinum neurotoxin type A (BoNT/A) has emerged as a viable option. However, its potential use in the management of TN is hampered by methodological limitations in existing studies and a lack of pivotal clinical trials. This study investigated the efficacy, optimal treatment site, preventive utility, and duration of effect of incobotulinumtoxinA (Inco/A), a BoNT/A, in a model of TN. Methods: An infraorbital nerve chronic constriction injury model was used to induce mechanical allodynia in male Sprague–Dawley rats, reproducing the trigeminal sensitization seen in TN. The effects of subcutaneous Inco/A (1, 2, and 4U) were measured using the mechanical sensitivity (von Frey) test to evaluate the dose response, effect of injection location, potential preventive nature of treatment, and duration of benefit. Results: Inco/A produced a robust, dose-dependent reduction in mechanical allodynia, predominantly via a local mechanism of action. Both preventive and therapeutic administration of Inco/A was efficacious, with significant reduction of allodynia even when administered up to 28 days before nerve injury. The anti-allodynic effect persisted up to 56 days post-injection. Conclusions: Inco/A is highly effective in alleviating mechanical allodynia in a validated rat model of TN. The findings highlight Inco/A as a promising candidate for clinical translation in TN and related neuropathic pain syndromes and support systematic investigation in well-controlled human trials.

Abstract: Predicting profitable entry signals in Bitcoin markets remains challenging due to price volatility, the absence of fundamental valuation frameworks, and methodological pitfalls that are common in the literature. In this study, we evaluate five machine learning classifiers using a 37-feature hierarchical multi-timeframe pipeline with price-level-agnostic normalisation across four temporal resolutions (15-minute, 4-hour, daily, and 3-day), spanning January 2020 to November 2025. Binary training labels were generated via majority-vote aggregation across 54 stop-loss/take-profit combinations, producing 6,951 balanced samples (48.5% positive class). Five algorithms --- Logistic Regression, Decision Tree, Random Forest, XGBoost, and LightGBM --- are compared using expanding-window TimeSeriesSplit validation (5 folds). Random Forest achieved the highest cross-validated ROC-AUC (0.6086), with all models showing modest but consistent discriminative ability (range 0.57--0.61). Feature importance analysis identifies 4-hour Bollinger Band position and RSI as dominant predictors, with all timeframes contributing meaningfully. A true out-of-sample holdout on 1,136 independently generated 2025 samples confirms generalisation, with Logistic Regression achieving 0.6087 ROC-AUC. A subtle multi-timeframe look-ahead bias in higher-timeframe data alignment is identified and corrected, which inflated performance by approximately 0.20 ROC-AUC points before correction. Event-driven backtesting on 2025 out-of-sample data yields a gross upper-bound return of +35.97% (185 trades, SL=1%, TP=2%, threshold=0.7, Sharpe=0.14) before transaction costs; after realistic round-trip fees, net returns are likely negligible. The central finding is that models with ROC-AUC $\approx$ 0.60 cannot reliably generate economically significant returns once transaction costs are accounted for. The methodology provides a reproducible framework for ML-based binary classification studies requiring transparent, bias-corrected validation across diverse market regimes.

Abstract: This study evaluates the potential of solar radiation management (SRM) to mitigate projected increases in rainfall and flood risk across four major urban centers in Eastern Africa. Flood dynamics under historical and future climate conditions were simulated using the Rainfall–Runoff–Inundation (RRI) model. Observed hydrological conditions were established using daily precipitation from the Climate Hazards Group Infrared Precipitation with Stations (CHIRPS), together with hydrological and topographic datasets. Future flood projections and associated impacts were derived from climate simulations produced by the Whole Atmosphere Community Climate Model version 6 (WACCM6) and solar climate intervention experiments from the Assessing Responses and Impacts of Solar climate intervention on the Earth system with Stratospheric Aerosol Injection (ARISE-SAI) framework, both forced by the Shared Socioeconomic Pathway SSP2-4.5. Model performance was evaluated using the Nash–Sutcliffe efficiency (NSE), coefficient of determination (R2), root mean square deviation (RMSD), and peak discharge error (PDE). The RRI model reproduced observed river discharge with reasonable skill, exhibiting lower RMSD and PDE values for the Ethiopian catchment compared to those in Kenya and Tanzania. Results indicate that SRM implemented through stratospheric aerosol injection (SAI) can reduce peak inundation depths and the spatial extent of flooding in selected flood-prone areas of Dar es Salaam and Addis Ababa. These reductions correspond to a decrease in the projected exposure of populations and critical infrastructure to flood hazards. While the findings suggest that albedo-based solar geoengineering may moderate flood impacts in some Eastern African cities, uncertainties remain, particularly in the representation of convective rainfall processes and the reliance on a single hydrological modeling framework in this study. Further research using improved climate simulations and ensemble-based hydrological approaches is recommended.

Abstract: Recent progress in text-to-music generation has enabled high-quality audio synthesis from natural language prompts. However, such models are at risk of unintended replication, raising concerns regarding originality and intellectual property. While training-time mitigation strategies can address this issue, they typically require retraining or curated datasets, limiting their practicality for largescale systems. Inference-time methods provide a more lightweight alternative but often involve a trade-off between fidelity and memorization risk. This work introduces Repulsive Guidance (RG), a systematic inference-time mitigation strategy that reduces memorization without disrupting the intended conditional guidance from the text prompt. RG operates by enforcing divergence between dual diffusion trajectories through a repulsive term applied only during early denoising steps, without reversing the conditional guidance from the prompt. Experiments on MusicBench with the TANGO model demonstrate that RG offers a complementary mitigation strategy, providing new insights into balancing fidelity and memorization risk.

Abstract: Based on the principal-agent theory, this paper constructs a n-level principal-agent model with multi-branches at the chain terminal, introduces the improved Fehr-Schmidt fairness preference theory, and establishes a n-level incentive framework covering the horizontal and vertical fairness preferences of terminal agents. By treating intermediate agents as a unified whole through the "black-box" approach, this paper focuses on analyzing the influence mechanism of terminal agents' fairness preferences on effort level, incentive contract design and the operational efficiency of the entire principal-agent chain. The research results confirm that incorporating fairness preferences into the incentive mechanism design of green supply chain principal-agent relationships can effectively stimulate the production and operation enthusiasm of terminal agents, improve their effort levels; at the same time, it can significantly reduce agency costs in the principal-agent chain, optimize resource allocation efficiency, and thereby enhance green production efficiency. The conclusions and model design of this study provide a new theoretical path and practical reference for the collaborative operation and sustainable development of actual green supply chains with terminal multi-branch principal-agent structures.

Abstract: Chaotic itinerancy—irregular switching among metastable collective states—provides a dynamical substrate for flexible social coordination, yet its mechanistic origin in multi-agent systems remains unclear. We present a multi-agent Active Inference model in which chaotic itinerancy emerges from Expected Free Energy minimization without outcome-level social priors. Agents select actions to minimize Expected Free Energy while updating preferences through a precision-gated learning mechanism modulated by interpersonal trust. Hill-function nonlinearity in state transitions creates bistable “affordance landscapes” that gate behavioral mode switching. Simulations with small number of agents on an Erdos–Rényi trust network reveal spontaneous alternation among multiple metastable behavioral clusters, heavy-tailed dwell-time distributions, and sign-changing finite-time Lyapunov exponents—three hallmarks of chaotic itinerancy. Crucially, replacing Hill-function dynamics with linear transitions reduces the chaotic-itinerancy detection rate from 80% to 20%, demonstrating that nonlinear affordance structure is necessary for generating metastable switching. We further show that agents with simplified internal models of the world sustain richer itinerant dynamics as a group than “perfect-foresight” agents, suggesting that bounded rationality may be functionally advantageous for maintaining behavioral flexibility. These results establish active inference as a principled framework for modeling chaotic itinerancy in social systems and offer a computational account of trust-mediated collective transitions observed in theatre workshops and group dynamics.

Abstract: This study examines the role of Enterprise Risk Management (ERM) in mitigating cyber fraud in Indonesian State-Owned Enterprises (SOEs). As digital transformation increases organizational exposure to cyber risks, effective risk governance mechanisms become essential for safeguarding financial integrity. This research investigates how ERM implementation contributes to cyber fraud prevention and detection within SOEs. The study employs a mixed-methods approach using quantitative panel data from 48 non-financial SOEs during the 2020–2024 period, resulting in 112 firm-year observations, complemented by qualitative insights from 25 key informants, including auditors, risk officers, and IT/cybersecurity specialists. The empirical analysis indicates that stronger ERM implementation significantly enhances firms’ ability to mitigate cyber fraud risks and improves coordination between financial risk management and information technology governance. The findings also highlight the importance of integrated risk governance structures in strengthening internal controls and organizational resilience against digital threats. This study contributes to the literature on risk governance and digital risk management by providing empirical evidence on the strategic role of ERM in enhancing financial accountability and cyber risk mitigation in emerging market SOEs.

Abstract: The act of receiving and giving help is commonly expected to improve older people’s health. In this article, this expectation is explored through a longitudinal analysis of a representative sample of 29,995 respondents aged 59 to 100 from 12 European countries documented in the Survey of Health, Ageing and Retirement in Europe (SHARE) for the period 2011 to 2022. An unordered correlated random-effects Mundlak (CRE), an ordered fixed-effects logistic (FEO), as well as an ordered random-effects Mundlak logistic panel estimator (REO), all with longitudinal calibrated weights, are used to estimate the relationship between self-perceived health and receiving and giving help. Additionally, the estimators are repeated for testing the robustness of the findings across the five available imputed SHARE datasets. The study supports the positive impact on health status of income, doing activities and being satisfied with life, as well as the negative impact of age, having limitations and being permanently sick. The Danish older people report the best health status, while the Swedish, Belgian, and Dutch jointly hold the second-best health status. Moreover, the value of a helping hand is revealed. Receiving help implies a 24.8 per cent increase in the odds of reporting worse health, yet, when combined with giving help, the latter lowers to 7.7 per cent. Overall, the study confirms the positive effect of receiving and giving help. This suggests that initiating a stimulus for recipients of help to engage in giving help is an opportunity to improve the self-perception of health among older people in Europe.

Abstract: Modern user authentication systems increasingly need user and device behavior aware adaptive mechanism to detect evolving threats beyond the traditional authentication framework of static credential verification. This paper proposes a hybrid multi-model framework for personalized user-level anomaly detection using a data-driven Hybrid Anomaly Score (HAS). Unlike static thresholding approaches, The proposed framework derives algorithm that integrates multiple anomaly detection methodologies to compute HAS through adaptive per-user thresholds (using cohort maturity and percentile-based optimization). The framework is evaluated on 72 million real-world data set. The framework demonstrate 96% precision, 92% recall, and an F1-score of 0.94, while maintaining inference latency within 2-3 ms per authentication event. The ablation analysis of the framework confirms the contribution of dynamic weighting and personalized threshold optimization to improved detection stability and convergence. The proposed framework outperforms existing approaches in both scalability and latency satisfying real-time operational constraint. The results indicate that data-driven adaptive thresholding combined with hybrid anomaly modeling provides an effective and deployable solution for large-scale authentication environments.

Abstract:

Summary: The potential of deep learning (DL) in genomic selection (GS) is constrained by the significant technical expertise required to design and implement neural networks. While DL has revolutionized fields like language processing and structural biology, its application in GS has not yet consistently outperformed traditional models like mixed linear models. The key to unlocking DL's power in GS lies in the exploration of network architectures tailored to genomic data, a process that demands intensive programming and poses a barrier for many researchers. To overcome this challenge, we developed Artificial Intelligence for Efficient and Versatile Evaluation and Representation (AI4EVER), a freely available graphical software platform that enables users to explore and apply machine learning (ML) models without any coding. AI4EVER integrates a graphical user interface (GUI) with a Python-based ML backend. The platform currently supports five models: Ridge Regression, Random Forest, Gradient Boosted Decision Trees, Multi-Layer Perceptron, and a customizable Keras-based neural network that can simultaneously predict multiple traits in a single model. A key feature of AI4EVER is optional incorporation of genome-wide association study (GWAS) results (p-values) as feature weights during model training, enabling biologically informed DL workflows. The platform further provides real-time visualization of model performance metrics and automated feature-importance outputs to enhance interpretability. AI4EVER also separates model training and prediction workflows, allowing trained models to be reused for independent prediction datasets. Using a representative maize dataset, we demonstrate that AI4EVER enables access to advanced AI, empowers genomic researchers to accelerate data-driven decision-making in breeding programs, ultimately lowering the barrier to artificial intelligence-enabled genetic improvement in crops and animals and human health management.