Multi-FusNet: fusion mapping of features for fine-grained image retrieval networks.

As the diversity and volume of images continue to grow, the demand for efficient fine-grained image retrieval has surged across numerous fields. However, the current deep learning-based approaches to fine-grained image retrieval often concentrate solely on the top-layer features, neglecting the relevant information carried in the middle layer, even though these information contains more fine-grained identification content. Moreover, these methods typically employ a uniform weighting strategy during hash code mapping, risking the loss of critical region mapping-an irreversible detriment to fine-grained retrieval tasks. To address the above problems, we propose a novel method for fine-grained image retrieval that leverage feature fusion and hash mapping techniques. Our approach harnesses a multi-level feature cascade, emphasizing not just top-layer but also intermediate-layer image features, and integrates a feature fusion module at each level to enhance the extraction of discriminative information. In addition, we introduce an agent self-attention architecture, marking its first application in this context, which steers the model to prioritize on long-range features, further avoiding the loss of critical regions of the mapping. Finally, our proposed model significantly outperforms existing state-of-the-art, improving the retrieval accuracy by an average of 40% for the 12-bit dataset, 22% for the 24-bit dataset, 16% for the 32-bit dataset, and 11% for the 48-bit dataset across five publicly available fine-grained datasets. We also validate the generalization ability and performance stability of our proposed method by another five datasets and statistical significance tests. Our code can be downloaded from https://github.com/BJFU-CS2012/MuiltNet.git.

SIG: Graph-Based Cancer Subtype Stratification With Gene Mutation Structural Information.

Somatic tumors have a high-dimensional, sparse, and small sample size nature, making cancer subtype stratification based on somatic genomic data a challenge. Current methods for improving cancer clustering performance focus on dimension reduction, integrating multi-omics data, or generating realistic samples, yet ignore the associations between mutated genes within the patient-gene matrix. We refer to these associations as gene mutation structural information, which implicitly includes cancer subtype information and can enhance subtype clustering. We introduce a novel method for cancer subtype clustering called SIG(Structural Information within Graph). As cancer is driven by a combination of genes, we establish associations between mutated genes within the same patient sample, pair by pair, and use a graph to represent them. An association between two mutated genes corresponds to an edge in the graph. We then merge these associations among all mutated genes to obtain a structural information graph, which enriches the gene network and improves its relevance to cancer clustering. We integrate the somatic tumor genome with the enriched gene network and propagate it to cluster patients with mutations in similar network regions. Our method achieves superior clustering performance compared to SOTA methods, as demonstrated by clustering experiments on ovarian and LUAD datasets. The code is available at https://github.com/ChangSIG/SIG.git.

CircGPC3 promotes hepatocellular carcinoma progression and metastasis by sponging miR-578 and regulating RAB7A/PSME3 expression.

CircRNAs are a class of highly stable noncoding RNAs that play an important role in the progression of many diseases, especially cancer. In this study, high-throughput sequencing was used to screen for abnormally expressed circRNAs, and we found that circGPC3 was overexpressed in HCC tissues. However, the underlying mechanism of circGPC3 in the development and metastasis of hepatocellular carcinoma (HCC) remains unknown. In our study, we found that circGPC3 was significantly upregulated in HCC tissues and cells and that its overexpression was positively correlated with overall survival, TNM stage and lymph node metastasis. In vivo and in vitro experiments showed that circGPC3 knockdown repressed HCC cell migration, invasion and proliferation and promoted apoptosis. Mechanistically, circGPC3 promoted HCC proliferation and metastasis through the miR-578/RAB7A/PSME3 axis. Our results demonstrate that circGPC3 contributes to the progression of HCC and provides an intervention target for HCC.

The Application of Artificial Intelligence and Big Data in the Food Industry.

Over the past few decades, the food industry has undergone revolutionary changes due to the impacts of globalization, technological advancements, and ever-evolving consumer demands. Artificial intelligence (AI) and big data have become pivotal in strengthening food safety, production, and marketing. With the continuous evolution of AI technology and big data analytics, the food industry is poised to embrace further changes and developmental opportunities. An increasing number of food enterprises will leverage AI and big data to enhance product quality, meet consumer needs, and propel the industry toward a more intelligent and sustainable future. This review delves into the applications of AI and big data in the food sector, examining their impacts on production, quality, safety, risk management, and consumer insights. Furthermore, the advent of Industry 4.0 applied to the food industry has brought to the fore technologies such as smart agriculture, robotic farming, drones, 3D printing, and digital twins; the food industry also faces challenges in smart production and sustainable development going forward. This review articulates the current state of AI and big data applications in the food industry, analyses the challenges encountered, and discusses viable solutions. Lastly, it outlines the future development trends in the food industry.

Genome-Wide Identification of PAP1 Direct Targets in Regulating Seed Anthocyanin Biosynthesis in Arabidopsis.

Anthocyanins are widespread water-soluble pigments in the plant kingdom. Anthocyanin accumulation is activated by the MYB-bHLH-WD40 (MBW) protein complex. In Arabidopsis, the R2R3-MYB transcription factor PAP1 activates anthocyanin biosynthesis. While prior research primarily focused on seedlings, seeds received limited attention. This study explores PAP1's genome-wide target genes in anthocyanin biosynthesis in seeds. Our findings confirm that PAP1 is a positive regulator of anthocyanin biosynthesis in Arabidopsis seeds. PAP1 significantly increased anthocyanin content in developing and mature seeds in Arabidopsis. Transcriptome analysis at 12 days after pollination reveals the upregulation of numerous genes involved in anthocyanin accumulation in 35S:PAP1 developing seeds. Chromatin immunoprecipitation and dual luciferase reporter assays demonstrate PAP1's direct promotion of ten key genes and indirect upregulation of TT8, TTG1, and eight key genes during seed maturation, thus enhancing seed anthocyanin accumulation. These findings enhance our understanding of PAP1's novel role in regulating anthocyanin accumulation in Arabidopsis seeds.

Prior Hospitalization, Severe Maternal Morbidity, and Pregnancy-Associated Deaths in Massachusetts From 2002 to 2019.

OBJECTIVE: To examine demographic and clinical precursors to pregnancy-associated deaths overall and when pregnancy-related deaths are excluded. METHODS: We conducted a retrospective cohort study based on a Massachusetts population-based data system linking data from live birth and fetal death certificates to corresponding delivery hospital discharge records and a birthing individual's nonbirth hospital contacts and associated death records. Exposures included maternal demographics, severe maternal morbidity (without transfusion), hospitalizations in the 3 years before pregnancy, comorbidities during pregnancy, and opioid use. In cases of postpartum deaths, hospitalization between delivery and death was examined. The primary outcome measure was pregnancy-associated death , defined as death during pregnancy or up to 1 year postpartum. RESULTS: There were 1,291,626 deliveries between 2002 and 2019, of which 384 were linked to pregnancy-associated deaths. Pregnancy-associated but not pregnancy-related deaths (per 100,000 deliveries) were highest for birthing people with opioid use before pregnancy (498.3), severe maternal morbidity (387.3), a comorbidity (106.3), or a prior hospitalization (88.9). In multivariable analysis, the adjusted risk ratios associated with severe maternal morbidity (9.37, 95% CI, 6.14-14.31) and opioid use (6.49, 95%, CI, 3.71-11.35) were highest. Individuals with pregnancy-associated deaths were also more likely to have been hospitalized before or during pregnancy (2.30, 95% CI, 1.62-3.26). Among postpartum deaths, more than two-thirds (69.9%) of birthing people had a hospital contact after delivery and before their death. CONCLUSION: Severe maternal morbidity and opioid use disorder were precursors to pregnancy-associated deaths. Individuals with pregnancy-associated but not pregnancy-related deaths experienced a history of hospital contacts during and after pregnancy before death.

Preconception and perinatal hospitalizations as indicators of risk for severe maternal morbidity in primiparas.

BACKGROUND: Severe maternal morbidity includes unexpected outcomes of labor and delivery that result in significant short- or long-term consequences to a woman's health. A statewide longitudinally linked database was used to examine hospitalization during and before pregnancy for birthing people with severe maternal morbidity at delivery. OBJECTIVE: This study aimed to examine the association between hospital visits during pregnancy and 1 to 5 years before pregnancy and severe maternal morbidity at delivery. STUDY DESIGN: This study was a retrospective, population-based cohort analysis of the Massachusetts Pregnancy to Early Life Longitudinal database between January 1, 2004, and December 31, 2018. Nonbirth hospital visits, including emergency department visits, observational stays, and hospital admissions during pregnancy and 5 years before pregnancy, were identified. The diagnoses for hospitalizations were categorized. We compared medical conditions leading to antecedent, nonbirth hospital visits among primiparous birthing individuals with singleton births with and without severe maternal morbidity, excluding transfusions. RESULTS: Of 235,398 birthing individuals, 2120 had severe maternal morbidity, a rate of 90.1 cases per 10,000 deliveries, and 233,278 did not have severe maternal morbidity. Compared with 4.3% of patients without severe maternal morbidity, 10.4% of patients with severe maternal morbidity were hospitalized during pregnancy. In multivariable analysis, there was a 31% increased risk of hospital admission during the prenatal period, a 60% increased risk of hospital admission in the year before pregnancy, and a 41% increased risk of hospital admission in 2 to 5 years before pregnancy. Compared with 9.8% of non-Hispanic White birthing people, 14.9% of non-Hispanic Black birthing people with severe maternal morbidity experienced a hospital admission during pregnancy. For those with severe maternal morbidity, prenatal hospitalization was most common for those with endocrine (3.6%) or hematologic (3.3%) conditions, with the largest differences between those with and without severe maternal morbidity for musculoskeletal (relative risk, 9.82; 95% confidence interval, 7.06-13.64) and cardiovascular (relative risk, 9.73; 95% confidence interval, 7.26-13.03) conditions. CONCLUSION: This study found a strong association between previous nonbirth hospitalizations and the likelihood of severe maternal morbidity at delivery.

A case report of Vibrio vulnificus sepsis in a diabetic patient.

Vibrio vulnificus is a facultative anaerobic, alkalophilic, halophilic, mesophilic, Gram-negative bacterium that can cause severe wound infection, sepsis and diarrhea. This paper reported a case of 85-year-old male patient infected with Vibrio vulnificus due to being stabbed by a sea shrimp. This patient also had diabetes with a long history of alcoholism. Due to bacterial pathogenicity and the patient's underlying diseases, his condition deteriorated rapidly. Based on the rapid diagnosis of Vibrio vulnificus using the next-generation sequencing（NGS）technology and blood culture method, as well as the selection of the most effective antibiotics via drug sensitivity test, this patient underwent precise antimicrobial treatment, thorough debridement and drainage within the shortest possible time, and thus the prognosis of this patient was greatly improved. In this paper, we have systematically explored the epidemiology, clinical features, diagnosis and treatment of Vibrio vulnificus infection, thus providing a practical reference for the clinicians to quickly identify and treat possible Vibrio vulnificus infection in diabetic patients after contacting with sea water or seafood.

Identifying Influential Nodes in Complex Networks Based on Information Entropy and Relationship Strength.

Identifying influential nodes is a key research topic in complex networks, and there have been many studies based on complex networks to explore the influence of nodes. Graph neural networks (GNNs) have emerged as a prominent deep learning architecture, capable of efficiently aggregating node information and discerning node influence. However, existing graph neural networks often ignore the strength of the relationships between nodes when aggregating information about neighboring nodes. In complex networks, neighboring nodes often do not have the same influence on the target node, so the existing graph neural network methods are not effective. In addition, the diversity of complex networks also makes it difficult to adapt node features with a single attribute to different types of networks. To address the above problems, the paper constructs node input features using information entropy combined with the node degree value and the average degree of the neighbor, and proposes a simple and effective graph neural network model. The model obtains the strength of the relationships between nodes by considering the degree of neighborhood overlap, and uses this as the basis for message passing, thereby effectively aggregating information about nodes and their neighborhoods. Experiments are conducted on 12 real networks, using the SIR model to verify the effectiveness of the model with the benchmark method. The experimental results show that the model can identify the influence of nodes in complex networks more effectively.

AFG1-induced TNF-α-mediated inflammation enhances gastric epithelial cell injury via CYP2E1.

Aflatoxin G1 (AFG1), a member of the aflatoxin family with cytotoxic and carcinogenic properties, is one of the most common mycotoxins occurring in various agricultural products, animal feed, and human foods and drinks worldwide. Epithelial cells in the gastrointestinal tract are the first line of defense against ingested mycotoxins. However, the toxicity of AFG1 to gastric epithelial cells (GECs) remains unclear. In this study, we explored whether and how AFG1-induced gastric inflammation regulates cytochrome P450 to contribute to DNA damage in GECs. Oral administration of AFG1 induced gastric inflammation and DNA damage in mouse GECs associated with P450 2E1 (CYP2E1) upregulation. Treatment with the soluble TNF-α receptor sTNFR:Fc inhibited AFG1-induced gastric inflammation, and reversed CYP2E1 upregulation and DNA damage in mouse GECs. TNF-α-mediated inflammation plays an important role in AFG1-induced gastric cell damage. Using the human gastric cell line GES-1, AFG1 upregulated CYP2E1 through NF-κB, causing oxidative DNA damage in vitro. The cells were also treated with TNF-α and AFG1 to mimic AFG1-induced TNF-α-mediated inflammation. TNF-α activated the NF-κB/CYP2E1 pathway to promote AFG1 activation, which enhanced DNA cellular damage in vitro. In conclusion, AFG1 ingestion induces TNF-α-mediated gastric inflammation, which upregulates CYP2E1 to promote AFG1-induced DNA damage in GECs.

Application of Three-Dimensional Digital Photogrammetry to Quantify the Surface Roughness of Milk Powder.

The surface appearance of milk powders is a crucial quality property since the roughness of the milk powder determines its functional properties, and especially the purchaser perception of the milk powder. Unfortunately, powder produced from similar spray dryers, or even the same dryer but in different seasons, produces powder with a wide variety of surface roughness. To date, professional panelists are used to quantify this subtle visual metric, which is time-consuming and subjective. Consequently, developing a fast, robust, and repeatable surface appearance classification method is essential. This study proposes a three-dimensional digital photogrammetry technique for quantifying the surface roughness of milk powders. A contour slice analysis and frequency analysis of the deviations were performed on the three-dimensional models to classify the surface roughness of milk powder samples. The result shows that the contours for smooth-surface samples are more circular than those for rough-surface samples, and the smooth-surface samples had a low standard deviation; thus, milk powder samples with the smoother surface have lower Q (the energy of the signal) values. Lastly, the performance of the nonlinear support vector machine (SVM) model demonstrated that the technique proposed in this study is a practicable alternative technique for classifying the surface roughness of milk powders.

ELAA: An Ensemble-Learning-Based Adversarial Attack Targeting Image-Classification Model.

The research on image-classification-adversarial attacks is crucial in the realm of artificial intelligence (AI) security. Most of the image-classification-adversarial attack methods are for white-box settings, demanding target model gradients and network architectures, which is less practical when facing real-world cases. However, black-box adversarial attacks immune to the above limitations and reinforcement learning (RL) seem to be a feasible solution to explore an optimized evasion policy. Unfortunately, existing RL-based works perform worse than expected in the attack success rate. In light of these challenges, we propose an ensemble-learning-based adversarial attack (ELAA) targeting image-classification models which aggregate and optimize multiple reinforcement learning (RL) base learners, which further reveals the vulnerabilities of learning-based image-classification models. Experimental results show that the attack success rate for the ensemble model is about 35% higher than for a single model. The attack success rate of ELAA is 15% higher than those of the baseline methods.

Advances in High-Performance Non-Ferrous Materials.

Non-ferrous metallic materials are considered to be fundamental materials for manufacturing in-dustries, i [...].

Neutral amino acid transporter SLC38A2 protects renal medulla from hyperosmolarity-induced ferroptosis.

Hyperosmolarity of the renal medulla is essential for urine concentration and water homeostasis. However, how renal medullary collecting duct (MCD) cells survive and function under harsh hyperosmotic stress remains unclear. Using RNA-Seq, we identified SLC38A2 as a novel osmoresponsive neutral amino acid transporter in MCD cells. Hyperosmotic stress-induced cell death in MCD cells occurred mainly via ferroptosis, and it was significantly attenuated by SLC38A2 overexpression but worsened by Slc38a2-gene deletion or silencing. Mechanistic studies revealed that the osmoprotective effect of SLC38A2 is dependent on the activation of mTORC1. Moreover, an in vivo study demonstrated that Slc38a2-knockout mice exhibited significantly increased medullary ferroptosis following water restriction. Collectively, these findings reveal that Slc38a2 is an important osmoresponsive gene in the renal medulla and provide novel insights into the critical role of SLC38A2 in protecting MCD cells from hyperosmolarity-induced ferroptosis via the mTORC1 signalling pathway.

DCU-Net: a dual-channel U-shaped network for image splicing forgery detection.

The detection and location of image splicing forgery are a challenging task in the field of image forensics. It is to study whether an image contains a suspicious tampered area pasted from another image. In this paper, we propose a new image tamper location method based on dual-channel U-Net, that is, DCU-Net. The detection framework based on DCU-Net is mainly divided into three parts: encoder, feature fusion, and decoder. Firstly, high-pass filters are used to extract the residual of the tampered image and generate the residual image, which contains the edge information of the tampered area. Secondly, a dual-channel encoding network model is constructed. The input of the model is the original tampered image and the tampered residual image. Then, the deep features extracted from the dual-channel encoding network are fused for the first time, and then the tampered features with different granularity are extracted by dilation convolution, and then, the secondary fusion is carried out. Finally, the fused feature map is input into the decoder, and the predicted image is decoded layer by layer. The experimental results on Casia2.0 and Columbia datasets show that DCU-Net performs better than the latest algorithm and can accurately locate tampered areas. In addition, the attack experiments show that DCU-Net model has good robustness and can resist noise and JPEG recompression attacks.

IFL-GAN: Improved Federated Learning Generative Adversarial Network With Maximum Mean Discrepancy Model Aggregation.

The generative adversarial network (GAN) is usually built from the centralized, independent identically distributed (i.i.d.) training data to generate realistic-like instances. In real-world applications, however, the data may be distributed over multiple clients and hard to be gathered due to bandwidth, departmental coordination, or storage concerns. Although existing works, such as federated learning GAN (FL-GAN), adopt different distributed strategies to train GAN models, there are still limitations when data are distributed in a non-i.i.d. manner. These studies suffer from convergence difficulty, producing generated data with low quality. Fortunately, we found that these challenges are often due to the use of a federated averaging strategy to aggregate local GAN models' updates. In this article, we propose an alternative approach to tackling this problem, which learns a globally shared GAN model by aggregating locally trained generators' updates with maximum mean discrepancy (MMD). In this way, we term our approach improved FL-GAN (IFL-GAN). The MMD score helps each local GAN hold different weights, making the global GAN in IFL-GAN getting converged more rapidly than federated averaging. Extensive experiments on MNIST, CIFAR10, and SVHN datasets demonstrate the significant improvement of our IFL-GAN in both achieving the highest inception score and producing high-quality instances.

Trends and inequities in severe maternal morbidity in Massachusetts: A closer look at the last two decades.

It is estimated that 50,000-60,000 pregnant people in the United States (US) experience severe maternal morbidity (SMM). SMM includes life-threatening conditions, such as acute myocardial infarction, acute renal failure, amniotic fluid embolism, disseminated intravascular coagulation, or sepsis. Prior research has identified both rising rates through 2014 and wide racial disparities in SMM. While reducing maternal death and SMM has been a global goal for the past several decades, limited progress has been made in the US in achieving this goal. Our objectives were to examine SMM trends from 1998-2018 to identify factors contributing to the persistent and rising rates of SMM by race/ethnicity and describe the Black non-Hispanic/White non-Hispanic rate ratio for each SMM condition. We used a population-based data system that links delivery records to their corresponding hospital discharge records to identify SMM rates (excluding transfusion) per 10, 000 deliveries and examined the trends by race/ethnicity. We then conducted stratified analyses separately for Black and White birthing people. While the rates of SMM during the same periods steadily increased for all racial/ethnic groups, Black birthing people experienced the greatest absolute increase compared to any other race/ethnic group going from 69.4 in 1998-2000 to 173.7 per 10,000 deliveries in 2016-2018. In addition, we found that Black birthing people had higher rates for every individual condition compared to White birthing people, with rate ratios ranging from a low of 1.11 for heart failure during surgery to a high of 102.4 for sickle cell anemia. Obesity was not significantly associated with SMM among Black birthing people but was associated with SMM among White birthing people [aRR 1.18 (95% CI: 1.02, 1.36)]. An unbiased understanding of how SMM has affected different race/ethnicity groups is key to improving maternal health and preventing SMM and mortality among Black birthing people. SMM needs to be addressed as both a medical and public health challenge.

Fixation-related fMRI analysis reveals the neural basis of parafoveal processing in self-paced reading of Chinese words.

While parafoveal word processing plays an important role in natural reading, the underlying neural mechanism remains unclear. The present study investigated the neural basis of parafoveal processing during Chinese word reading with the co-registration of eye-tracking and functional magnetic resonance imaging (fMRI) using fixation-related fMRI analysis. In the gaze-contingent boundary paradigm, preview conditions (words that are identical, orthographically similar, and unrelated to target words), pre-target word frequency and target word frequency were manipulated. When fixating the pre-target word, the identical preview condition elicited lower brain activation in the left fusiform gyrus relative to unrelated and orthographically similar preview conditions and there were significant interactions of preview condition and pre-target word frequency on brain activation of the left middle frontal gyrus, left fusiform gyrus and supplementary motor area. When fixating the target word, there was a significant main effect of preview condition on brain activation of the right fusiform gyrus and a significant interaction of preview condition and pre-target word frequency on brain activation of the left middle frontal gyrus. These results suggest that fixation-related brain activation provides immediate measures and new perspectives to understand the mechanism of parafoveal processing in self-paced reading.

GFCNet: Utilizing graph feature collection networks for coronavirus knowledge graph embeddings.

In response to fighting COVID-19 pandemic, researchers in machine learning and artificial intelligence have constructed some medical knowledge graphs (KG) based on existing COVID-19 datasets, however, these KGs contain a considerable amount of semantic relations which are incomplete or missing. In this paper, we focus on the task of knowledge graph embedding (KGE), which serves an important solution to infer the missing relations. In the past, there have been a collection of knowledge graph embedding models with different scoring functions to learn entity and relation embeddings published. However, these models share the same problems of rarely taking important features of KG like attribute features, other than relation triples, into account, while dealing with the heterogeneous, complex and incomplete COVID-19 medical data. To address the above issue, we propose a graph feature collection network (GFCNet) for COVID-19 KGE task, which considers both neighbor and attribute features in KGs. The extensive experiments conducted on the COVID-19 drug KG dataset show promising results and prove the effectiveness and efficiency of our proposed model. In addition, we also explain the future directions of deepening the study on COVID-19 KGE task.

Length of Stay Among Infants with Neonatal Abstinence Syndrome and Risk of Hospital Readmission.

OBJECTIVES: To assess whether a shorter length of stay (LOS) is associated with a higher risk of readmission among newborns with neonatal abstinence syndrome (NAS) and examine the risk, causes, and characteristics associated with readmissions among newborns with NAS, using a longitudinally linked population-based database. METHODS: Our study sample included full-term singletons with NAS (n = 4,547) and without NAS (n = 327,836), born in Massachusetts during 2011-2017. We used log-binomial regression models to estimate the crude risk ratios (cRRs) and adjusted RRs with 95% confidence intervals (CI) of the association between LOS and readmissions, controlling for maternal age, race/ethnicity, education, marital status, insurance, method of delivery, birthweight, adequacy of prenatal care, smoking, and abnormal conditions of newborn. RESULTS: Compared with infants without NAS, infants with NAS had a non-significantly higher risk of readmission within 2-42 days (2.8% vs. 2.5%; p = 0.17) and a significantly higher risk of readmission within 43-182 days (2.7% vs. 1.8%; p < 0.001). The risk of readmission within 2-42 days was significantly higher among infants with NAS with a LOS of 0-6 days compared to a LOS of 14-20 days (reference group) (aRR: 2.1; 95% CI: 1.2-3.5). No significant differences in readmission rates between 43 and 182 days were observed across LOS categories. CONCLUSIONS: Among infants with NAS, a LOS of 0-6 days was associated with a significantly higher risk of readmission within 2-42 days of discharge compared to a longer LOS.