Browse Articles

Privacy-preserving cyberthreat detection in decentralized social media with federated cross-modal graph transformers

Abstract The new era of decentralized, privacy-oriented social media platforms has brought us a set of related enforcement problems which include detecting cyberbullying, disinformation on a coordinated scale 5,14 . These centralized or unimodal systems are unable to work efficiently when faced with stringent privacy concerns or multimodal content. In this paper, we present Federated Cross-Modal Graph Transformer (FCMGT) to jointly model text, image and audio features and social graph structure in federated learning settings. Furthermore, the proposed approach is enhanced by a dynamic adversarial training to mitigate content perturbation, graph manipulation and model-poisoning attacks. On a large-scale synthetic decentralized dataset (2 M + interactions), the experiments reveal that FCMGT achieves an F1-Score of 0.927, outperforming the best baseline by 4.6%, and achieves an AUC of 0.963. Performance drop down under adversarial attacks is only 3.8%, in contrast to 15–30% for previous models. These findings position FCMGT as a reliable, scalable, and privacy-preserving system for safe guarding next-generation decentralized social networks.

Barriers and facilitators to establishing Early Psychosis Intervention (EPI) services in remote and rural communities across Canada

Aims There is a paucity of data on how Early Psychosis Intervention (EPI) services adapt to the challenges of rural practice, including geographical isolation and provider shortages. This qualitative study seeks to fill this gap by exploring the experiences of EPI programs in rural and remote communities in Canada. Methods An introductory email was sent to 117 EPI contacts, inviting them to first complete a survey and then participate in a semi-structured interview if their program contained a rural component. The interviews were conducted between June 2023-January 2024 via Zoom or phone by a member of our research team, while another member took detailed notes. Qualitative data was analyzed using thematic analysis. Results Twenty-five representatives from seventeen distinct EPI programs with a rural and remote component across seven Canadian provinces and one Canadian territory participated in the interviews. Several barriers to establishing EPI programs in rural and remote areas were identified, including limited access to care, challenges in referral processes, geographical and technological constraints, and funding limitations. Facilitators of successful program delivery included the use of virtual care, standardized care models, strong leadership and teamwork, and adaptive resourcefulness in addressing local needs. Conclusion Identifying and addressing these barriers and leveraging facilitators can enhance the accessibility and effectiveness of EPI service delivery in rural and remote areas.

KidneyTox_v1.0 enables explainable artificial intelligence prediction of nephrotoxicity in small molecules

Abstract Drug-induced nephrotoxicity remains a leading cause of kidney dysfunction, often with severe or even fatal outcomes. Computational approaches, in particular artificial intelligence (AI), offer a promising alternative by providing reliable, cost-effective, and ethically sound tools for assessing drug-induced nephrotoxicity. Thereby, potentially reducing reliance on animal testing. This study was driven by three core objectives: (i) to analyze the chemical space of compounds associated with drug-induced nephrotoxicity, (ii) to construct a robust supervised machine learning (ML) model for classification, followed by a quantitative Read-Across Structure-Activity Relationship (qRASAR) study, and (iii) to develop an open-access, eXplainable AI (XAI) platform named “KidneyTox_v1.0” ( https://kidneytoxv1.streamlit.app/ ) for nephrotoxicity prediction. Beyond providing predictions, “KidneyTox_v1.0” offers interpretability through interactive SHAP-based waterfall plots, enabling both domain experts and non-experts to understand the contribution of molecular descriptors to toxicity outcomes. These modelling analyses will assist chemists in designing less nephrotoxic molecules in the future.

Correction: Correction: A machine learning approach for estimating forage maize yield and quality in NW Spain

Associations between edentulism and risk of neurodegenerative diseases among middle-aged and older adults in China: a decade-long cohort study

Consumer preferences for the development of new emulsion products based on vegetable and animal fats

The purpose of this paper was to determine consumer preferences for the acceptance of a new emulsion product containing interesterified fat based on mutton tallow and hemp oil. The survey concerned consumers’ reference to the new emulsion products as pharmaceutical, food, and cosmetic products. On the basis of the survey, it was concluded that the hypothetical emulsion products would gain consumer acceptance. Acquisition of this knowledge forms the basis for making decisions regarding the implementation of further experimental work, the outcome of which will be the evaluation of the properties of prototype products, including a comparative assessment of the sensory characteristics of prototypes and market products. More than half of the respondents answered that they would be interested in a new emulsion product based on interesterified fat. The most important factors indicated by the respondents that would lead them to purchase cosmetic or pharmaceutical emulsions were improvement in skin condition after application and the possibility of testing a free sample in advance. In the case of a food emulsion containing modified mutton tallow and hemp oil, on the other hand, the most important factors determining a potential purchase were the opportunity to try a free sample of the product in advance and attributes such as taste and texture that corresponded to the respondents.

A multi-branch network for cooperative spectrum sensing via attention-based and CNN feature fusion

Exploring the effects of arch support height on badminton sidestep cutting using musculoskeletal modeling

Sidestep cutting is a fundamental yet high-risk movement in badminton, often placing considerable biomechanical stress on the lower limbs. Arch support insoles are commonly used to improve foot stability and redistribute loads, but their specific biomechanical effects during badminton movements remain unclear. This study investigated how different arch support heights influence lower limb mechanics during a 45-degree sidestep cutting (45C) maneuver in badminton players, with a focus on joint kinematics, kinetics, and contact forces. Fifteen male athletes performed 45C maneuver trials under three insole conditions: no support (NS), low support (LS), and high support (HS). Lower limb biomechanics were analyzed using musculoskeletal modeling in OpenSim, assessing joint angles, joint moments, and joint contact forces at the ankle, knee, and hip. A one-way repeated-measures ANOVA was used to evaluate differences across conditions. The high-support condition significantly increased ankle dorsiflexion (P = 0.002) and knee external rotation (P < 0.001) compared with the no-support and low-support conditions. Hip extension moments were higher under the high-support condition (P = 0.043). Anterior–posterior ankle joint contact force increased under high support (P = 0.014), while no significant differences were observed in knee or hip joint contact forces across conditions. These findings describe acute mechanical responses under controlled laboratory conditions. Whether these responses relate to long-term movement strategies in different athlete populations remains to be clarified in future work.

The impact of digital literacy on green consumption behavior under the moderation of prosocial behavior

Research on road parametric modeling and dynamic lightweighting methods driven by BIM-GIS integration

Addressing the challenges of inflexibility and low modeling efficiency in the forward design of road building information modeling (BIM), as well as the performance bottlenecks within integrated BIM-GIS environments, this study proposes a novel integration-driven method for parametric modeling and dynamic lightweight processing of roads (PMDL). The proposed approach integrates terrain-adaptive algorithms with lightweight rendering techniques, thereby enabling rapid design iteration and dynamic optimization. Based on this method, a parametric system for 3D road model features is constructed according to the spatial topological relationships between the road and the terrain. Shape grammar is employed to drive the road pavement modeling process, ensuring both design flexibility and modeling efficiency. For complex terrain conditions, a threshold-triggered dynamic modeling mechanism is designed. Through terrain elevation analysis, the undulation of road sections is automatically identified. Algorithms are developed for continuous road pavement generation, Grid-based segmentation of the slopes, dynamic calculation of pier heights, and automatic tunnel generation, enabling the adaptive creation of roads, bridges, tunnels, and slopes. Finally, quadric error metrics (QEM) mesh simplification, level of detail (LOD), and view frustum culling are applied to optimize the loading efficiency and rendering performance of 3D models on the OpenSceneGraphEarth (OSGEarth) platform. This achieves a stable frame rate above 50 FPS for large-scale scenes, effectively resolving rendering lag issues in large-scale scenarios. Experiments show that compared to traditional oblique photography modeling (OPM) and differential elements method (DiEM), this method significantly improves modeling speed, accuracy, and data scheduling efficiency, providing efficient technical support for intelligent design, dynamic updating, and multi-scale visualization of digital twin roads.

Green analytical evaluation of anticancer drug analysis: A multi-tool assessment of HPLC and LC–MS methods

Laminin N-terminus α31 regulates corneal epithelial cell adhesion and migration through modifying the organization and proteolytic processing of laminin 332

Laminin N-terminus α31 (LaNt α31) is a netrin-like protein generated by alternative splicing of the laminin α3 gene. While previously shown to regulate vascular permeability in vivo, its role in epithelial tissues remains less defined. Here, we demonstrate that LaNt α31 modulates epithelial cell behavior by altering laminin 332 (LM332) organization and hemidesmosome (HD) maturation. Adenoviral driven overexpression of LaNt α31 in corneal epithelial cells led to premature HD assembly, marked by enhanced recruitment of collagen XVII and BPAG1e to β4 integrin, and reduced cell migration. LaNt α31 expression reorganized LM332 from diffuse arcs into tight clusters and co-localized with LMβ3 during matrix deposition. Notably, phenotypes were rescued by precoated extracellular matrix, indicating a matrix-dependent mechanism. Furthermore, LaNt α31 increased matrix metalloproteinase (MMP) activity and LMα3 proteolytic processing, both essential for its effects, as MMP inhibition reversed LM clustering and HD maturation. These findings identify LaNt α31 as a regulator of epithelial homeostasis through modulation of LM332 architecture and cell–matrix adhesion.

Identification of stable resistant wild pigeonpea donors to multiple predominant bruchid species in India

Exploring collaborative strategies to improve patient safety in healthcare organizations: A qualitative study

Background Patient safety remains a critical concern in healthcare, necessitating the exploration of collaborative approaches to enhance care quality and outcomes. This qualitative study delves into the significance of interprofessional collaboration, leadership support, patient engagement, and safety culture in driving patient safety initiatives within healthcare settings. Method This qualitative phenomenological approach, involving in-depth interviews and document analysis, was conducted to explore the perspectives of healthcare professionals, administrators, patients, policymakers, and researchers involved in patient safety initiatives. Thematic analysis was employed to identify and organize recurring patterns in the data, while interpretive phenomenological analysis was utilized to gain a deeper understanding of participants’ lived experiences. This dual approach revealed key themes related to teamwork, communication, leadership, patient engagement, and best practices in patient safety. Results The study revealed that interprofessional collaboration is widely recognized as vital for patient safety, with participants highlighting the importance of effective communication, shared decision-making, and mutual respect among healthcare professionals. Leadership support, organizational structures, and a culture of safety emerged as key enabling factors. Despite the benefits, challenges such as hierarchy, siloed communication, resource constraints, and resistance to change remain significant barriers. Strategies including regular interdisciplinary training, patient engagement initiatives, leveraging technology, and continuous quality improvement processes were identified as effective ways to strengthen collaborative practices. Conclusion Promoting interprofessional collaboration, strong leadership, and a robust safety culture are essential for improving patient safety outcomes. Overcoming systemic and interpersonal barriers requires targeted interventions at both individual and organizational levels, with ongoing education, patient participation, and data-driven feedback playing crucial roles in sustaining collaborative efforts. These insights support the adoption of integrated approaches to foster safer and higher-quality care across healthcare settings.

From chessboard of bipolarons of size 4a in cubic $$\hbox {La}_{7/8}\hbox {Sr}_{1/8}\hbox {MnO}_3$$ to stripes of the same bipolarons in layered high $$T_c$$ cuprates

The effects of vaginal gel from Myrtus communis on the sexual function of married women during reproductive aging: A study protocol for a randomized controlled trial

Introduction Human sexuality is an important aspect of functionality, and many patients believe that it determines their quality of life. Sexual dysfunction can lead to stress, strained relationships, and a decrease in self-esteem. The majority of modern medical treatments for improving female sexual function are associated with significant side effects and high costs. Traditionally, M. communis has been used to treat sexual impotence. Here, we present the protocol of an interventional clinical phase II study to test the hypothesis that vaginal gel containing Myrtus communis extract can improve sexual function in women of reproductive age. Methods/Design In a prospective, randomized, placebo-controlled and highly blind clinical phase II trial, 80 women aged 18--40 years with sexual dysfunction meeting the inclusion criteria will be randomized to an intervention group receiving a vaginal gel containing myrtle leaf extract (n = 38) or an active control group receiving a placebo gel with an identical appearance (n = 38). Randomization will be performed via a permuted block technique with random allocation software. This study will be conducted at the women’s clinic of Imam Reza Hospital in Mashhad. Blinded assessments of outcome variables will be conducted twice: before treatment and one month after treatment. The primary outcome measure was sexual function. Discussion This randomized controlled clinical trial evaluates the efficacy of the vaginal gel Myrtus communis on the sexual function of married women during reproductive age. The study design presented here fulfills the criteria of a high-quality clinical phase II trial of sexual function. Trial registration IRCT.behdasht.gov.ir Identifier: IRCT20230723058892N1

ONCOPLEX: an oncology-inspired hypergraph model integrating diverse biological knowledge for cancer driver gene prediction

Abstract Cancer development is driven by a small subset of somatic mutations, known as driver mutations, that disrupt key regulatory processes in cells. These mutations occur in specific genes, called cancer driver genes, whose altered functions promote tumor initiation and progression. Accurately identifying driver genes remains a major challenge due to their rarity and the overwhelming presence of passenger mutations. Recent advances in graph-based deep learning have improved the modeling of gene interactions, but most approaches are limited to pairwise connections and fail to capture the higher-order complexity of biological systems. We introduce ONCOPLEX, a hypergraph-based neural network framework that models genes as nodes and curated cancer-related pathways as hyperedges, enabling the representation of multi-gene interactions. Unlike previous methods, ONCOPLEX integrates diverse molecular and phenotypic features, such as somatic mutations, gene expression, and DNA methylation, into a pathway-informed hypergraph structure to learn biologically meaningful gene representations. ONCOPLEX is trained in a supervised manner on labeled driver and non-driver genes, with unlabeled genes included as nodes during representation learning. Comprehensive evaluations across pan-cancer and cancer-type-specific settings show that ONCOPLEX consistently outperforms state-of-the-art methods in classification and ranking metrics. It accurately recovers known driver genes and highlights novel candidates supported by literature and enrichment analyses. These findings underscore the power of pathway-guided hypergraph modeling for advancing cancer driver gene discovery.

The association between smoking and clinical outcomes among spondylodesis patients: A systematic review and meta-analysis

Background The effects of smoking on outcomes after spinal spondylodesis remain unclear due to conflicting findings in the existing literature. This meta-analysis aimed to evaluate the association between smoking and clinical as well as radiological outcomes following instrumented spinal fusion, and to examine differences based on smoking status, spinal region, and surgical extent. Methods PubMed, Embase, and Cochrane were searched up to August 2024 for comparative studies reporting outcomes in smokers and non-smokers undergoing spinal fusion. Extracted outcomes included clinical scores (Oswestry Disability Index [ODI], Neck Disability Index [NDI], Visual Analog Scale [VAS] for pain), fusion rates, pseudoarthrosis, and complications. Pooled percent mean change from baseline and weighted incidences with 95% confidence intervals (CI) were calculated using a random-effects model. Subgroup analyses were performed by spinal region, number of operated levels, and smoking history (current, former, never). Results Twenty-nine studies involving 6,687 patients were included. In most outcome measures, smokers showed smaller percent improvements than non-smokers. For instance, NDI improved by 42.8% (95% CI: 25.4–60.1%) in smokers vs. 49.7% (32.4–64.0%) in non-smokers; ODI improved by 48.6% (34.7–62.5%) vs. 56.5% (42.6–70.3%); and VAS back pain by 55.0% (29.7–80.2%) vs. 60.5% (35.2–85.7%). Fusion rates were lower in smokers (86.8%) than in non-smokers (95.1%), while pseudoarthrosis was more common in smokers (17.2% vs. 7.3%). Subgroup analyses revealed similar trends across spinal regions and surgical scope. A hierarchical pattern was observed, with never-smokers experiencing the most favorable outcomes, followed by former smokers. Conclusion In most instances, smokers appeared to experience worse outcomes following instrumented spinal procedures compared to non-smokers, and a hierarchical pattern was notable, with current smokers experiencing the worst outcomes, followed by former smokers compared to never-smokers. Future well-designed studies with proper adjustment for confounding are needed to further confirm these findings.

Network accessibility as the emergence of cliques

Investigating the antioxidant, antibacterial, and antifungal properties of typha domingensis leaves assisted synthesized zinc oxide nanoparticles

Green-synthesized nanoparticles (NPs) offer unique properties over traditional approaches, with applications in coatings, catalysts, packaging, environmental cleaning, and medicine. This research involves the synthesis of Zinc oxide nanoparticles (ZnO NPs) utilizing aqueous leaf extract of Typha domingensis (locally known as Hogla pata) as a reducing agent. The biosynthesized NPs were then characterized using ultraviolet-visible spectroscopy (UV-Vis), Fourier transform infrared (FTIR) spectroscopy, field emission scanning electron microscopy (FESEM), and energy-dispersive X-ray (EDX) spectroscopy. The XRD result revealed that the synthesized NPs were enriched with (101) facets, which were similar to the standard JCPDS card no 00-036-1451. The size of the ZnO NPs was determined from XRD data utilizing the Debye-Scherrer equation and found to be around 26.74 nm. Moreover, the SEM analysis showed that the shape of the NPs was agglomerated spherical. The 2,2-diphenyl-1-picrylhydrazyl (DPPH) free radical showed dose-dependent antioxidant activity against synthesized ZnO NPs, with an average IC 50 of 32.25 µg/mL. The antibacterial efficacy of green-synthesized ZnO NPs was assessed against fungal and bacterial pathogens in antifungal experiments. Colletotrichum sp. showed the highest sensitivity to ZnO NPs, with a 20 ± 0.80 mm zone of inhibition (ZOI) and 22.22% inhibition, whereas Fusarium sp. had a 17 ± 0.68 mm ZOI and 18.89% inhibition at a dosage of 2 mg/mL. ZnO NPs also exhibited antibacterial activity against two gram-negative bacteria, Coliform sp. and Salmonella sp. , both demonstrating a 12 mm ZOI and 13.33% inhibition at 5.0 mg/mL. These results demonstrate that ZnO NPs biosynthesized with Typha domingensis leaf extract exhibit versatile antimicrobial and antioxidant properties, offering a safe and sustainable alternative to synthetic compounds.