Explore the vast range of titles available.

Visual question answering (VQA) is a multi-modal task involving natural language processing (NLP) and computer vision (CV), which requires models to understand of both visual information and textual information simultaneously to predict the correct answer for the input visual image and textual question, and has been widely used in smart and intelligent transport systems, smart city, and other fields. Today, advanced VQA approaches model dense interactions between image regions and question words by designing co-attention mechanisms to achieve better accuracy. However, modeling interactions between each image region and each question word will force the model to calculate irrelevant information, thus causing the model’s attention to be distracted. In this paper, to solve this problem, we propose a novel model called Multi-modal Explicit Sparse Attention Networks (MESAN), which concentrates the model’s attention by explicitly selecting the parts of the input features that are the most relevant to answering the input question. We consider that this method based on top-k selection can reduce the interference caused by irrelevant information and ultimately help the model to achieve better performance. The experimental results on the benchmark dataset VQA v2 demonstrate the effectiveness of our model. Our best single model delivers 70.71% and 71.08% overall accuracy on the test-dev and test-std sets, respectively. In addition, we also demonstrate that our model can obtain better attended features than other advanced models through attention visualization. Our work proves that the models with sparse attention mechanisms can also achieve competitive results on VQA datasets. We hope that it can promote the development of VQA models and the application of artificial intelligence (AI) technology related to VQA in various aspects.

Vision-language models aim to seamlessly integrate visual and linguistic information for multi-modal tasks, demanding precise semantic alignments between image-text pairs while minimizing the influence of irrelevant data. While existing methods leverage intra-modal and cross-modal knowledge to enhance alignments, they often fall short in sufficiently reducing interference, which ultimately constrains model performance. To address this gap, we propose a novel vision-language model, the threshold-based knowledge integration network (TBKIN), designed to effectively capture intra-modal and cross-modal knowledge while systematically mitigating the impact of extraneous information. TBKIN employs unified scene graph structures and advanced masking strategies to strengthen semantic alignments and introduces a fine-tuning strategy based on threshold selection to eliminate noise. Comprehensive experimental evaluations demonstrate the efficacy of TBKIN, with our best model achieving state-of-the-art accuracy of 73.90% on the VQA 2.0 dataset and 84.60% on the RefCOCO dataset. Attention visualization and detailed result analysis further validate the robustness of TBKIN in tackling vision-language tasks. The model’s ability to reduce interference while enhancing semantic alignments underscores its potential for advancing multi-modal learning. Extensive experiments across four widely-used benchmark datasets confirm its superior performance on two typical vision-language tasks, offering a practical and effective solution for real-world applications.

Background and objective: Sacituzumab govitecan (SG), the first antibody-drug conjugate targeting human trophoblast cell-surface antigen 2 (Trop-2), has been approved by the Food and Drug Administration (FDA) for the treatment of advanced or metastatic breast cancer and urothelial cancer. However, there is currently a dearth of information regarding the safety profiles of SG in a large sample cohort. The objective of the present study is to investigate SG-related adverse events (AEs) in real-world settings leveraging the FDA Adverse Event Reporting System (FAERS) database to guide the safety management of clinical medication. Methods: The FAERS database was retrospectively queried to extract reports associated with SG from April 2020 to March 2023. To identify and evaluate potential AEs in patients receiving SG, various disproportionality analyses such as reporting odds ratio (ROR), the proportional reporting ratio (PRR), the Bayesian confidence propagation neural network (BCPNN), and the multi-item gamma Poisson shrinker (MGPS) were employed. Results: Overall, 2069 reports of SG as the “primary suspect” were identified. Noteworthy, SG was significantly associated with an increased risk of blood lymphatic system disorders (ROR, 7.18; 95% CI, 6.58–7.84) and hepatobiliary disorders (ROR, 2.68; 95% CI, 2.17–3.30) at the System Organ Class (SOC) level. Meanwhile, 61 significant disproportionality preferred terms (PTs) simultaneously complied with all four algorithms were adopted. Therein, anemia, thrombocytopenia, neutropenia, leukopenia, diarrhea, asthenia, alopecia, and electrolyte imbalance were consistent with the common AEs described in the clinical trials and specification of SG. Furthermore, unexpected significant AEs include colitis (ROR, 12.09; 95% CI, 9.1–16.08), heart rate increased (ROR, 5.11; 95% CI, 3.84–6.79), sepsis (ROR, 4.77; 95% CI, 3.59–6.34), cholestasis (ROR, 6.28; 95% CI, 3.48–11.36), blood bilirubin increased (ROR, 4.65; 95% CI, 2.42–8.94) and meningitis (ROR, 7.23; 95% CI, 2.71–19.29) were also be detected. The median time to onset of SG-related AEs was 14 [interquartile range (IQR), 7–52] days, with the majority occurring within the initial month of SG treatment. Conclusion: Our study validates the commonly known AEs and also found some potentially emerging safety issues related to SG in real-world clinical practice, which could provide valuable vigilance evidence for clinicians and pharmacists to manage the safety issues of SG.

As important mobile elements in prokaryotes, prophages shape the genomic context of their hosts and regulate the structure of bacterial populations. However, it is challenging to precisely identify prophages through computational methods. Here, we introduce PIDE for identifying prophages from bacterial genomes or metagenome-assembled genomes. PIDE integrates a pre-trained protein language model and gene density clustering algorithm to distinguish prophages. Benchmarking with induced prophage sequencing datasets demonstrates that PIDE pinpoints prophages with precise boundaries. Applying PIDE to 4744 human gut representative genomes reveals 24,467 prophages with widespread functional capacity. PIDE is available at https://github.com/chyghy/PIDE , with model training code at https://zenodo.org/records/16457629 .

Advances in Artificial Intelligence envision a promising future, where the personalized Internet of Things can be revolutionized with the ability to continuously improve system efficiency and service quality. However, with the introduction of laws and regulations about data security and privacy protection, centralized solutions, which require data to be collected and processed directly on a central server, become impractical for personalized Internet of Things to train Artificial Intelligence models for a variety of domain-specific scenarios. Motivated by this, this paper introduces Cedar, a secure, cost-efficient and domain-adaptive framework to train personalized models in a crowdsourcing-based and privacy-preserving manner. In essentials, Cedar integrates federated learning and meta-learning to enable a safeguarded knowledge transfer within personalized Internet of Things for models with high generalizability that can be rapidly adapted by individuals. Through evaluation using standard datasets from various domains, Cedar is seen to achieve significant improvements in saving, elevating, accelerating and enhancing the learning cost, efficiency, speed, and security, respectively. These results reveal the feasibility and robust-ness of federated meta-learning in orchestrating heterogeneous resources in the cloud-edge-device continuum and defending malicious attacks commonly existed in the Internet, thereby unlockingthe potential of Artificial Intelligence in reforming personalized Internet of Things. This article addresses the challenge of training AI models in personalized Internet of Things systems while ensuring data security and privacy. Authors combine federated and meta-learning to improve model training performance, adaptation speed, cost-efficiency, and security against attacks.

School bullying significantly impacts adolescent physical and mental development. The current study aimed to explore the effect of campus exclusion on school bullying behavior among junior high school students and the role of callous-unemotional traits and family caring. The Campus Exclusion Questionnaire, Olweus Child Bullying Questionnaire, Callous-Unemotional Trait Scale, and Family Caring Scale were completed by 705 students. A moderated mediation model was analyzed using SPSS 24.0. Results indicated that both campus exclusion and callous-unemotional traits positively predicted bullying behavior. Callous-unemotional traits partially mediated the relationship between campus exclusion and bullying behavior. Additionally, family caring moderated the link between callous-unemotional traits and bullying behavior, mitigating adverse effects. The study highlighted family caring’s protective role against bullying linked to adverse school experiences. Therefore, collaboration between schools and families is crucial to reduce bullying.

Given the rapid development of social media, effective extraction and analysis of the contents of social media for health care have attracted widespread attention from health care providers. As far as we know, most of the reviews focus on the application of social media, and there is a lack of reviews that integrate the methods for analyzing social media information for health care. This scoping review aims to answer the following 4 questions: (1) What types of research have been used to investigate social media for health care, (2) what methods have been used to analyze the existing health information on social media, (3) what indicators should be applied to collect and evaluate the characteristics of methods for analyzing the contents of social media for health care, and (4) what are the current problems and development directions of methods used to analyze the contents of social media for health care? A scoping review following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines was conducted. We searched PubMed, the Web of Science, EMBASE, the Cumulative Index to Nursing and Allied Health Literature, and the Cochrane Library for the period from 2010 to May 2023 for primary studies focusing on social media and health care. Two independent reviewers screened eligible studies against inclusion criteria. A narrative synthesis of the included studies was conducted. Of 16,161 identified citations, 134 (0.8%) studies were included in this review. These included 67 (50.0%) qualitative designs, 43 (32.1%) quantitative designs, and 24 (17.9%) mixed methods designs. The applied research methods were classified based on the following aspects: (1) manual analysis methods (content analysis methodology, grounded theory, ethnography, classification analysis, thematic analysis, and scoring tables) and computer-aided analysis methods (latent Dirichlet allocation, support vector machine, probabilistic clustering, image analysis, topic modeling, sentiment analysis, and other natural language processing technologies), (2) categories of research contents, and (3) health care areas (health practice, health services, and health education). Based on an extensive literature review, we investigated the methods for analyzing the contents of social media for health care to determine the main applications, differences, trends, and existing problems. We also discussed the implications for the future. Traditional content analysis is still the mainstream method for analyzing social media content, and future research may be combined with big data research. With the progress of computers, mobile phones, smartwatches, and other smart devices, social media information sources will become more diversified. Future research can combine new sources, such as pictures, videos, and physiological signals, with online social networking to adapt to the development trend of the internet. More medical information talents need to be trained in the future to better solve the problem of network information analysis. Overall, this scoping review can be useful for a large audience that includes researchers entering the field.

Mental health plays a key role across the cancer care continuum, from prognosis and active treatment to survivorship and palliative care. Digital health technologies offer an appealing, cost-effective tool to address psychological needs. This umbrella review aims to summarize and evaluate the available evidence on the efficacy of digital health interventions for improving mental health and psychosocial outcomes for populations with cancer. Literature searches were conducted in Embase, PsycINFO, PubMed, CINAHL, the Cochrane Library, and Web of Science from their inception to February 4, 2024. Systematic reviews (with or without meta-analysis) investigating the efficacy of digital health interventions for psychosocial variables in patients with cancer were included. Quality was assessed using the Assessing the Methodological Quality of Systematic Reviews-2 tool. In total, 78 systematic reviews were included in this review. Among diverse delivery modalities and types of digital interventions, websites and smartphone apps were the most commonly used. Depression was the most frequently addressed, followed by quality of life, anxiety, fatigue, and distress. The qualities of the reviews ranged from critically low to high. Generally, despite great heterogeneity in the strength and credibility of the evidence, digital health interventions were shown to be effective for mental health in patients with cancer. Taken together, digital health interventions show benefits for patients with cancer in improving mental health. Various gaps were identified, such as little research specifically focusing on older adult patients with cancer, a scarcity of reporting high-precision emotion management, and insufficient attention to other certain mood indicators. Further exploration of studies with standardized and rigorous approaches is required to inform practice. PROSPERO CRD42024565084; https://tinyurl.com/4cbxjeh9.

TGF-β1 can induce epithelial-mesenchymal transition (EMT) in primary airway epithelial cells (AECs). Semaphorin7A (Sema7a) plays a crucial role in regulating immune responses and initiating and maintaining transforming growth factor β1 TGF-β1-induced fibrosis. To determine the expression of Sema7a, in serum isolated from asthmatics and non-asthmatics, the role of Sema7a in TGF-β1 induced proliferation, migration and airway EMT in human bronchial epithelial cells (HBECs) . The concentrations of Sema7a in serum of asthmatic patients was detected by enzyme-linked immunosorbent assay (ELISA). The expressions of Sema7a and integrin-β1 were examined using conventional western blotting and real-time quantitative PCR (RT-PCR). Interaction between the Sema7a and Integrin-β1 was detected using the Integrin-β1 blocking antibody (GLPG0187). The changes in EMT indicators were performed by western blotting and immunofluorescence, as well as the expression levels of phosphorylated Focal-adhesion kinase (FAK) and Extracellular-signal-regulated kinase1/2 (ERK1/2) were analyzed by western blot and their mRNA expression was determined by RT-PCR. We described the first differentially expressed protein of sema7a, in patients with diagnosed bronchial asthma were significantly higher than those of healthy persons (P<0.05). Western blotting and RT-PCR showed that Sema7a and Integrin-β1 expression were significantly increased in lung tissue from the ovalbumin (OVA)-induced asthma model. GLPG0187 inhibited TGF-β1-mediated HBECs EMT, proliferation and migration, which was associated with Focal-adhesion kinase (FAK) and Extracellular-signal-regulated kinase1/2 (ERK1/2) phosphorylation. Sema7a may play an important role in asthma airway remodeling by inducing EMT. Therefore, new therapeutic approaches for the treatment of chronic asthma, could be aided by the development of agents that target the Sema7a.

Background Gut microbiota and microbiota-derived metabolites have been implicated in the regulation of stress-related diseases, yet their associations with chronic stress in adolescents remain unclear. Multi-omics studies on this topic in adolescents are still limited. This study aimed to characterize gut microbiota and metabolites in adolescents under chronic stress. Methods In this cross-sectional study, we assessed chronic stress in 124 adolescents aged 12–16 years using the Adolescent Life Events Scale and the Study Stress Scale. Participants were stratified by stress level into low ( n  = 42), medium ( n  = 41), and high stress ( n  = 41) groups. Fecal samples were collected from all participants for 16S rRNA gene sequencing. Subsequently, a subset of 30 adolescents with high stress and 29 low stress adolescents underwent metagenomic sequencing and untargeted metabolomics. Results Adolescents experiencing high-chronic stress showed lower alpha diversity, differential beta diversity, and a more complicated microbial network compared to those experiencing lower stress. Spearman’s rank correlation and Kruskal-Wallis test identified five genera with decreased abundances in high stress adolescents, including Faecalibacterium , Bacteroides , Akkermansia , Lachnospiraceae unclassified , and Ruminococcus ( P fdr <0.05). Additionally, 12 species showed decreased abundances and 5 increased abundances, and logistic regression analysis further revealed that the relative abundances of Bifidobacterium catenulatum , Streptococcus suis , Ruminococcus sp. CAG 108 , and Phascolarctobacterium faecium were associated with chronic stress ( P fdr <0.05), after adjusting for sex, age, fruit consumption, and body mass index. We identified 21 differential metabolites, predominantly enriched in metabolic pathways based on KEGG analysis. Moreover, 19 out of these metabolites were significantly correlated with at least one of the four species significantly associated with chronic stress. These metabolites may explain health effects of species associated with chronic stress. Conclusions Chronic stress in adolescents is associated with altered gut microbiota composition and metabolite profiles, providing insights into possible mechanisms underlying stress-related diseases and highlighting the importance of longitudinal studies to clarify temporal and causal relationships.