Advancing Open Science

Visual Relationship Prediction (VRP) is crucial for advancing structured scene understanding, yet existing methods struggle with ineffective multimodal fusion, static relationship representations, and a lack of logical consistency. To address these limitations, this paper proposes a Hierarchical CLIP model (H-CLIP) for structured scene parsing. Our approach leverages a pre-trained CLIP backbone to extract aligned visual, textual, and spatial features for entities and their union regions. A multi-head self-attention mechanism then performs deep, dynamic multimodal fusion. The core innovation is a consistency and reversibility verification mechanism, which imposes algebraic constraints as a regularization loss to enforce logical coherence in the learned relation space. Extensive experiments on the Visual Genome dataset demonstrate the superiority of the proposed method. H-CLIP significantly outperforms state-of-the-art baselines on the predicate classification task, achieving a Recall@50 score of 64.31% and a Mean Recall@50 of 36.02%, thereby validating its effectiveness in generating accurate and logically consistent scene graphs even under long-tailed distributions.

Objectives: Free-living amoeba Naegleria fowleri causes primary amoebic meningoencephalitis (PAM). While infection is rare, PAM’s fatality rate exceeds 97%. The recommended treatment includes combination therapy, which does not result in uniform survival. Thus, there is a critical unmet need for finding better therapy for PAM. Drug repurposing can expedite the discovery of effective treatment for PAM. Isavuconazonium is approved for the treatment of fungal infections. Given that isavuconazole is the major metabolite of isavuconazonium and isavuconazole penetrates into the brain with high efficiency, our objective was to determine the activity of both isavuconazonium and isavuconazole on N. fowleri trophozoites. Methods: To test the effect of both compounds, we determined their dose–responses against N. fowleri and two mammalian cells. To establish how fast the prodrug and the metabolite kill the trophozoites, we measured potency at different time points. Finally, we investigated the effect of combining isavuconazonium or isavuconazole with amphotericin B on both N. fowleri and mammalian cells. Results: Both isavuconazonium and the metabolite isavuconazole were active against multiple strains, with clinically relevant isavuconazole exhibiting potency ranging between 0.1 and 0.6 µM. They were less toxic on mammalian cells. Isavuconazonium and isavuconazole required 24 h to achieve nanomolar potency. Combination with amphotericin B was synergistic without eliciting toxicity on mammalian cells. Conclusions: Our findings, together with the use of intravenous and oral formulations of isavuconazonium to treat pediatric and adult patients, support further in vivo efficacy study of isavuconazonium for its potential use for the treatment of PAM.

Oxidative stress is a key driver of chronic inflammatory diseases. Anodendron affine is a native Formosan plant species in Taiwan that remains largely underexplored phytochemically and bioactivity. To reveal the bioactive constituents and assess its potential as a source of anti-inflammatory antioxidants, we performed bioactivity-guided fractionation and evaluated the inhibition of superoxide anion (O₂^•–) generation in formyl-L-methionyl-L-leucyl-L-phenylalanine-stimulated human neutrophils. Molecular docking simulations were employed to model interactions with Formyl peptide receptor 1 (FPR1) and the Nicotinamide adenine dinucleotide phosphate (NADPH) oxidase complex, including neutrophil cytosol factor 1 (p47phox) and NADPH oxidase 2 (NOX2), to propose a theoretical mechanism of action. Phytochemical investigation led to the isolation of two new compounds, methyl 4,5-O-diferuloyl-3-methoxyquinate (1) and 16-pregnen-3,12,20-trione (2), together with four known compounds. Notably, 4-hydroxy-3-prenylbenzoic acid (5) exhibited potent inhibitory activity (IC₅₀ = 17.65 ± 0.97 μM), surpassing the activity of the positive control, ibuprofen (IC₅₀ = 27.85 ± 3.56 μM). Docking studies suggested that anodendrosin H (4) and 4-hydroxy-3-prenylbenzoic acid (5) exhibit high predicted binding affinity to p47phox and NOX2. Based on these results, compounds 1, 4, and 5 from A. affine were identified as potential lead candidates for the development of novel anti-inflammatory therapeutics.

Signal transducer and activator of transcription 2 (STAT2) is a key component of the type I interferon (IFN-I/III) signaling pathway, which is pivotal in host defense against cancer and viral infections and in shaping immune responses. Building on our previously reported conditional Stat2 knockout (KO) mouse, we expand its utility by validating additional tissue-specific models and exploring novel functional contexts. Mice carrying loxP-flanked Stat2 alleles were crossed with CMV-Cre, Cdx2-Cre or CD11c-Cre mice. Deletion of STAT2 was validated by PCR genotyping and western blotting in the relevant tissues. To confirm defective IFN-I signaling with STAT2 deletion, IFN-β stimulation of splenocytes from CMV-Cre Stat2 KO mice showed a lack of induction of canonical IFN-I target genes, confirming functional disruption of the pathway. In vivo, global Stat2 deletion significantly impaired the antitumor efficacy of IFN-β treatment. Similarly, lung fibroblasts isolated from globally deleted Stat2 KO mice showed defective antiviral responses to IFN-β. Tissue-specific Cre models demonstrated selective ablation of STAT2 in target compartments without affecting its expression in non-target tissues. Together, these studies expand our published conditional Stat2 KO findings and highlight the value of this model as a versatile platform for dissecting STAT2-dependent signaling pathways in a tissue- and disease-specific manner.

Data uncertainty and limited model generalization remain critical bottlenecks in large-scale remote sensing of soil salinization. Although the integration of multi-source data has improved predictive potential, conventional deterministic feature selection methods often overlook stochastic noise inherent in environmental variables, leading to models that overfit spurious correlations rather than learning stable physical signals. To address this limitation, this study proposes a Bootstrap–Boruta feature stability assessment framework that shifts feature selection from deterministic “feature importance” ranking to probabilistic “feature stability” evaluation, explicitly accounting for uncertainty induced by data perturbations. The proposed framework is evaluated by integrating stability-driven feature sets with multiple machine learning models, including a Back-Propagation Neural Network (BPNN) optimized using the Red-billed Blue Magpie Optimization (RBMO) algorithm as a representative optimization strategy. Using the Minqin Lake region as a case study, the results demonstrate that the stability-based framework effectively filters unstable noise features, reduces systematic estimation bias, and improves predictive robustness across different modeling approaches. Among the tested models, the RBMO-optimized BPNN achieved the highest accuracy. Under a rigorous bootstrap validation framework, the quality-controlled ensemble model yielded a robust mean R² of 0.657 ± 0.05 and an RMSE of 1.957 ± 0.289 dS/m. The framework further identifies eleven physically robust predictors, confirming the dominant diagnostic role of shortwave infrared (SWIR) indices in arid saline environments. Spatial mapping based on these stable features reveals that 30.7% of the study area is affected by varying degrees of soil salinization. Overall, this study provides a mechanism-driven, promising, within-region framework that enhances the reliability of remote-sensing-based soil salinity inversion under heterogeneous environmental conditions.

Heat shock transcription factors (HSFs) play a central role in mediating plant responses to abiotic stress. Anthocyanins, one of the most important secondary metabolites in plants, contribute to both stress tolerance and the enhancement in crop nutritional quality. However, the possible role of HSFs in regulating anthocyanin biosynthesis in sweet potato (Ipomoea batatas L.) remains unknown. This study conducted a genome-wide analysis of the sweet potato HSF gene family to explore their functions related to anthocyanin metabolism and salinity stress. Multiple stress-inducible promoter elements were identified within IbHSF22, including those induced by drought, salt, heat, ABA, and light. For functional characterization of this gene, a 35S-driven overexpression construct was prepared and then transformed into Nicotiana benthamiana. Overexpression of IbHSF22 led to a nearly two-fold increase in anthocyanin content, concurrently with the elevated expression of key structural genes such as NtCHS, NtF3H, NtDFR, and NtANS. Under salt stress, the transgenic plants also exhibited enhanced tolerance, which was associated with maintained antioxidant enzyme activity and concerted induction of stress-responsive genes, events that collectively resulted in decreased oxidative damage. Therefore, the present work identifies IbHSF22 as an integrator of anthocyanin biosynthesis and salt defense mechanisms. These findings provide a conceptual basis and candidate gene strategy for dual improvement in stress resilience and nutritional quality in sweet potato breeding.

Background/Objectives: Effective nutritional care is a critical component of oncology treatment. Behaviour change techniques (BCTs) are key elements that drive individual behaviour change and are commonly identified within clinical interventions and implementation strategies. Despite their relevance, the application of BCTs in oncology nutrition has not been systematically examined. This systematic review aimed to identify and synthesise the use of BCTs in nutrition interventions and their implementation strategies within adult oncology care. Methods: A search across 10 electronic databases was conducted from inception up to December 2024. Data extraction focused on intervention characteristics, plus implementation barriers, strategies, and outcomes, which were coded using relevant established taxonomies, i.e., Theoretical Domains Framework, Behaviour Change Techniques, and Proctor’s outcomes framework. Study quality was assessed using the QuADS tool. Results: After screening 4055 abstracts and 165 full texts, 38 articles representing 31 unique studies were included. Nutrition interventions (n = 31) spanned across risk screening/assessment, care planning/pathways, nutritional monitoring, and support guidelines. Among the 19 interventions that incorporated BCTs targeting patients and/or healthcare professionals (HCPs), 20 unique BCTs were identified across 97 instances. Separately, implementation strategies reported in 25 of the 38 articles included 96 BCT instances (18 unique) targeting HCPs. BCTs were frequently observed alongside positive outcomes across patient, service, and implementation levels. Conclusions: Application of BCTs in oncology nutrition practice remains implicit and inconsistent. For clinical practice, more explicit specifications of BCTs may support better integration of nutrition care in routine oncology settings. Future research is warranted to test the effectiveness of specific BCTs and their combinations. This can be supported by standardised reporting of intervention content and implementation strategies which can enable identification of patterns of success and optimise replicability.

Background: This study investigates the global burden of early-onset bladder cancer (EOBC) from 1990 to 2021, highlighting regional disparities and the growing role of metabolic risk factors. Early-onset bladder cancer (EOBC), diagnosed before age 50, is an emerging global health concern. While less common than kidney cancer, EOBC contributes substantially to mortality and disability-adjusted life years (DALYs), with marked sex disparities. Its global epidemiology remains unassessed systematically. Methods: Using GBD 1990–2021 data, we analyzed EOBC incidence, prevalence, mortality, and DALYs across 204 countries in individuals aged 15–49. Trends were examined via segmented regression, EAPC, and Bayesian age-period-cohort modeling. Inequality was quantified using SII and CI. Decomposition and SDI-efficiency frontier analyses were introduced. Results: From 1990 to 2021, EOBC incidence rose 62.2%, prevalence 73.1%, deaths 15.3%, and DALYs 15.8%. Middle-SDI regions bore the highest burden. Aging drove trends in high-SDI areas and population growth in low-SDI regions. Over 25% of high-SDI countries underperformed in incidence/prevalence control. Smoking remained the leading risk factor, with rising hyperglycemia burdens in high-income areas. Males carried over twice the female burden, peaking at age 45–49. Conclusions: EOBC shows sustained global growth with middle-aged concentration and significant regional disparities. Structural inefficiencies highlight the need for enhanced screening, early warning, and tailored resource allocation.