A survey on causal inference for recommendation.

Causal inference has recently garnered significant interest among recommender system (RS) researchers due to its ability to dissect cause-and-effect relationships and its broad applicability across multiple fields. It offers a framework to model the causality in RSs such as confounding effects and deal with counterfactual problems such as offline policy evaluation and data augmentation. Although there are already some valuable surveys on causal recommendations, they typically classify approaches based on the practical issues faced in RS, a classification that may disperse and fragment the unified causal theories. Considering RS researchers' unfamiliarity with causality, it is necessary yet challenging to comprehensively review relevant studies from a coherent causal theoretical perspective, thereby facilitating a deeper integration of causal inference in RS. This survey provides a systematic review of up-to-date papers in this area from a causal theory standpoint and traces the evolutionary development of RS methods within the same causal strategy. First, we introduce the fundamental concepts of causal inference as the basis of the following review. Subsequently, we propose a novel theory-driven taxonomy, categorizing existing methods based on the causal theory employed, namely those based on the potential outcome framework, the structural causal model, and general counterfactuals. The review then delves into the technical details of how existing methods apply causal inference to address particular recommender issues. Finally, we highlight some promising directions for future research in this field. Representative papers and open-source resources will be progressively available at https://github.com/Chrissie-Law/Causal-Inference-for-Recommendation.

SIRT1-Mediated HMGB1 Deacetylation Suppresses Neutrophil Extracellular Traps Related to Blood-Brain Barrier Impairment After Cerebral Venous Thrombosis.

Cerebral venous thrombosis (CVT) is a neurovascular disease with recently increasing incidence. Aseptic inflammatory responses play an important role in the pathology of CVT. Recent studies report that neutrophil extracellular traps (NETs) are major triggers of thrombosis and inflammation in stroke, but their effect on brain injury in CVT requires further validation. In this study, two CVT animal models were used to simulate superior sagittal sinus thrombosis and cortical vein thrombosis. The effects of brain tissue infiltration of NETs and the molecular mechanisms associated with NET formation were deeply explored in combination with proteomics, histology, and serology. The results showed that the cortical vein thrombosis model could be combined with more severe blood-brain barrier (BBB) disruption and showed more severe cerebral hemorrhage. Decreased Sirtuin 1 (SIRT1) expression promotes high mobility group box 1 (HMGB1) acetylation, causing increased cytosolic translocation and extracellular release, and HMGB1 can promote NET formation and recruitment. In addition, corticocerebral accumulation of NETs contributes to BBB damage. This establishes a vicious cycle between BBB damage and NET accumulation. SIRT1 mediated-HMGB1 deacetylation may play a critical role in attenuating BBB damage following CVT. This study employed a combined validation using models of venous sinus thrombosis and cortical vein thrombosis to investigate the deacetylation role of SIRT1, aiming to offer new insights into the pathological mechanisms of brain injury following CVT.

Effects of BPD tendencies and subjective well-being on NSSI in adolescents with PTSD.

Background: Severe posttraumatic stress disorder (PTSD) may lead to non-suicidal self-injury (NSSI), and borderline personality disorder (BPD) tendencies may play a role in this process. Secondary vocational students experience more social, familial and other pressures and are more vulnerable to psychological problems. Thus, we explored the effect of BPD tendencies and subjective well-being (SWB) on NSSI in secondary vocational students with PTSD. Methods: A total of 2,160 Chinese secondary vocational students in Wuhan participated in our cross-sectional investigation. The Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), criteria for PTSD, NSSI Questionnaire, Personality Diagnostic Questionnaire-4+, subjective well-being scale, and family adaptation, partnership, growth, affection, and resolve (APGAR) Index were used. We conducted a binary logistic regression model and linear regression analysis. Results: Sex (odds ratio [OR] = 0.354, 95% confidence interval [CI] = 0.171-0.733), BPD tendencies (OR = 1.192, 95% CI = 1.066-1.333) and SWB (OR = 0.652, 95% CI = 0.516-0.824) were independent factors that predicted NSSI in secondary vocational students with PTSD. Spearman's correlation analysis showed that BPD tendencies were positively correlated with NSSI frequency (r = 0.282, P < 0.01). SWB was negatively correlated with NSSI frequency (r = -0.301, P < 0.01). The linear regression showed that BPD tendencies (ß = 0.137, P < 0.05 and ß = -0.230, P < 0.001) were significantly correlated with NSSI frequency. Spearman's correlation analysis showed that family functioning was positively correlated with SWB (r = 0.486, P < 0.01) and negatively correlated with BPD tendencies (r = -0.296, P < 0.01). Conclusion: In adolescents, PTSD in response to stressful events could lead to NSSI, and BPD tendencies promote the intensity of NSSI, while SWB diminishes its intensity. Improvement in family functioning may actively guide the development of mental health and improve SWB; such steps may constitute interventions to prevent or treat NSSI.

Proteomics analyses of acute kidney injury biomarkers in a rat exertional heat stroke model.

The frequency of exertional heat stroke (EHS) increases with the gradual elevation of global temperatures during summer. Acute kidney injury (AKI) is a common complication of EHS, and its occurrence often indicates the worsening of a patient's condition or a poor prognosis. In this study, a rat model of AKI caused by EHS was established, and the reliability of the model was evaluated by HE staining and biochemical assays. The expression of kidney tissue proteins in the EHS rats was analyzed using label-free liquid chromatography-tandem mass spectrometry. A total of 3,129 differentially expressed proteins (DEPs) were obtained, and 10 key proteins were finally identified, which included three upregulated proteins (Ahsg, Bpgm, and Litaf) and seven downregulated proteins (medium-chain acyl-CoA synthetase 2 (Acsm2), Hadha, Keg1, Sh3glb1, Eif3d, Ambp, and Ddah2). The qPCR technique was used to validate these 10 potential biomarkers in rat kidney and urine. In addition, Acsm2 and Ahsg were double-validated by Western blotting. Overall, this study identified 10 reliable biomarkers that may provide potential targets for the treatment of AKI caused by EHS.

Online Knowledge Distillation via Mutual Contrastive Learning for Visual Recognition.

The teacher-free online Knowledge Distillation (KD) aims to train an ensemble of multiple student models collaboratively and distill knowledge from each other. Although existing online KD methods achieve desirable performance, they often focus on class probabilities as the core knowledge type, ignoring the valuable feature representational information. We present a Mutual Contrastive Learning (MCL) framework for online KD. The core idea of MCL is to perform mutual interaction and transfer of contrastive distributions among a cohort of networks in an online manner. Our MCL can aggregate cross-network embedding information and maximize the lower bound to the mutual information between two networks. This enables each network to learn extra contrastive knowledge from others, leading to better feature representations, thus improving the performance of visual recognition tasks. Beyond the final layer, we extend MCL to intermediate layers and perform an adaptive layer-matching mechanism trained by meta-optimization. Experiments on image classification and transfer learning to visual recognition tasks show that layer-wise MCL can lead to consistent performance gains against state-of-the-art online KD approaches. The superiority demonstrates that layer-wise MCL can guide the network to generate better feature representations. Our code is publicly avaliable at https://github.com/winycg/L-MCL.

How to prevent malicious use of intelligent unmanned swarms?

With advancements in swarm intelligence, artificial intelligence, and wireless mobile network technology, unmanned swarms such as unmanned aerial vehicles, ground vehicles, ships, and other unmanned systems are becoming increasingly autonomous and intelligent. Benefiting from these technologies, intelligent unmanned swarms are able to efficiently perform complex tasks through collaboration in various fields. However, malicious use of intelligent unmanned swarms raises concerns about the potential for significant damage to national infrastructures such as airports and power facilities. Defending against malicious activities is essential but challenging due to the swarms' abilities to perceive, understand complex environments, and make accurate decisions through multi-system collaboration. This perspective sheds light on recent research in counter-measures and provides new trends and insights on how to prevent malicious actions by intelligent unmanned swarms.

Pneumonia in Children During the 2019 Outbreak in Xiamen, China.

BACKGROUND: To understand the regional epidemiology and clinical characteristics of adenovirus pneumonia in hospitalized children during the 2019 outbreak of respiratory adenoviruses in China. METHODS: We analyzed the epidemiologic trend of adenovirus in children hospitalized for acute lower respiratory tract infections in Xiamen in 2019. Adenovirus was identified using direct fluorescent antibody detection. During the peak seasons of adenovirus epidemic, 170 adenovirus-positive specimens were obtained for molecular typing, and the clinical data were collected. RESULTS: Among the 9890 children hospitalized for acute lower respiratory tract infection, 609 (6.2%) were tested positive for adenovirus. The detection rate of adenovirus was significantly higher in boys than in grils (9.5% vs. 4.6%, P < 0.05). Adenovirus activity increased markedly between April and August with the prevalence of 7.3%-12.4%. During the outbreak season, type 7 accounted for 70.6%, followed by type 3 (28.8%) and type 4 (0.6%). Of the 155 cases of adenovirus pneumonia, the median age was 3.0 years (range: 4 month to 9 years), 153 (98.7%) had fever with a mean fever duration of 9.04 ± 5.52 days, 28 (16.5%) had wheezing, 93 (60%) showed segmental or lobar consolidation with atelectasis and 13 (8.4%) showed pleural effusion. Forty-six (29.6%) cases developed severe pneumonia, 7 (4.1%) required mechanical ventilation and 2 (1.2%) died. Younger age, longer duration of fever and higher fever spike were more frequently seen in severe cases (P < 0.05). Twenty-five (16.2%) had C-reactive protein ≥ 40 mg/L, and 91 (58.7%) had procalcitonin ≥ 0.25 mg/L. CONCLUSIONS: Adenovirus types 7 and 3 caused the outbreak of adenovirus pneumonia in community children during late spring to summer in 2019 in Xiamen. The majority of adenovirus pneumonia resembles bacterial pneumonia. The incidence of severe pneumonia was high when type 7 predominantly prevailed. Adenovirus type 7 was more common in severe cases than in nonsevere cases.

Knowledge Distillation Using Hierarchical Self-Supervision Augmented Distribution.

Knowledge distillation (KD) is an effective framework that aims to transfer meaningful information from a large teacher to a smaller student. Generally, KD often involves how to define and transfer knowledge. Previous KD methods often focus on mining various forms of knowledge, for example, feature maps and refined information. However, the knowledge is derived from the primary supervised task, and thus, is highly task-specific. Motivated by the recent success of self-supervised representation learning, we propose an auxiliary self-supervision augmented task to guide networks to learn more meaningful features. Therefore, we can derive soft self-supervision augmented distributions as richer dark knowledge from this task for KD. Unlike previous knowledge, this distribution encodes joint knowledge from supervised and self-supervised feature learning. Beyond knowledge exploration, we propose to append several auxiliary branches at various hidden layers, to fully take advantage of hierarchical feature maps. Each auxiliary branch is guided to learn self-supervision augmented tasks and distill this distribution from teacher to student. Overall, we call our KD method a hierarchical self-supervision augmented KD (HSSAKD). Experiments on standard image classification show that both offline and online HSSAKD achieves state-of-the-art performance in the field of KD. Further transfer experiments on object detection further verify that HSSAKD can guide the network to learn better features. The code is available at https://github.com/winycg/HSAKD.

Cerebral venous hemodynamic responses in a mouse model of traumatic brain injury.

Traumatic brain injury (TBI) is a serious public health problem that endangers human health and is divided into primary and secondary injuries. Previous work has confirmed that changes in cerebral blood flow (CBF) are related to the progression of secondary injury, although clinical studies have shown that CBF monitoring cannot fully and accurately evaluate disease progression. These studies have almost ignored the monitoring of venous blood flow; however, as an outflow channel of the cerebral circulation, it warrants discussion. To explore the regulation of venous blood flow after TBI, the present study established TBI mouse models of different severities, observed changes in cerebral venous blood flow by laser speckle flow imaging, and recorded intracranial pressure (ICP) after brain injury to evaluate the correlation between venous blood flow and ICP. Behavioral and histopathological assessments were performed after the intervention. The results showed that there was a significant negative correlation between ICP and venous blood flow (r = -0.795, P < 0.01), and both recovered to varying degrees in the later stages of observation. The blood flow changes in regional microvessels were similar to those in venous, and the expression of angiogenesis proteins around the impact area was significantly increased. In conclusion, this study based on the TBI mouse model, recorded the changes in venous blood flow and ICP and revealed that venous blood flow can be used as an indicator of the progression of secondary brain injury.

Clinical Analysis of the Discovery of Malignant Gynecological Tumors in the Diagnosis and Treatment of Pelvic Organ Prolapse.

Background: Clinically, malignant gynecological tumors found by chance during the diagnosis and treatment of pelvic organ prolapse (POP) are rare, and they are usually missed, leading to delayed diagnosis and treatment. The initial treatment of these tumors cannot be standardized, and, as a single surgical intervention may not be able to treat both the tumor and prolapse, secondary surgery is usually needed, affecting the quality of life of patients. Case presentation: The present study retrospectively analyzed the data of three patients who were diagnosed with malignant gynecological tumors during the diagnosis and treatment of POP. These patients were among 215 patients with POP treated in Yuncheng Central Hospital of Shanxi Province between January 2011 and May 2020. The case characteristics, surgical interventions, postoperative treatments, and follow-ups were summarized, and the characteristics of diagnosis and treatment were analyzed in the context of relevant literature. Conclusion: As long as clinicians operate in strict accordance with the standards of diagnosis and treatment, obtain a complete medical history, undertake a physical examination, and remain diligent in auxiliary examinations, following existing clinical methods and diagnosis and treatment processes, patients with POP complicated with malignant gynecological tumors can be clearly diagnosed before and during surgery. In this way, initial treatment can be standardized, and surgical methods can be selected that address both the tumor and prolapse, thereby avoiding secondary surgery and improving the patient's quality of life.

Predictive Factors for Early-Onset Seizures in Patients With Cerebral Venous Sinus Thrombosis.

Seizures are reported to be important factors contributing to poor prognosis in patients with cerebral venous sinus thrombosis (CVST). However, the predictive factors for concurrent early onset seizures in patients with CVST remain unclear. To identify the predictive factors of early seizures in patients with CVST, this study retrospectively evaluated the clinical data of patients diagnosed with CVST at two centers from January 2011 to December 2020 and analyzed the relationship between admission characteristics and early onset seizures. A total of 112 CVST patients (63 men and 49 women; mean age 39.82 ± 15.70 years) were enrolled in this study, of whom 34 (30.36%) had seizures. For patients with seizures, cerebral hemorrhage, cortical vein thrombosis, anterior superior sagittal sinus (SSS) thrombosis, middle SSS thrombosis, CVST score, modified Rankin Scale, National Institute of Health Stroke Scale (NIHSS) score, neutrophil percentage, and D-dimer level were more severe than those without seizures. Logistic regression analysis showed that cerebral hemorrhage (P = 0.002), anterior SSS thrombosis (P = 0.003), NIHSS score ≥5 (P = 0.003), and D-dimer ≥0.88 mg/L (P = 0.004) were all significant predictive factors of early-onset seizures in CVST patients. Combining the four factors further improved the predictive capability with an area under the curve of 0.871 (95% confidence interval = 0.803-0.939). Further large-scale prospective studies are required to confirm these findings.

A novel rat model for cerebral venous sinus thrombosis: verification of similarity to human disease via clinical analysis and experimental validation.

BACKGROUND: Cerebral venous sinus thrombosis (CVST) is a rare neurovascular disorder with highly variable manifestations and clinical courses. Animal models properly matched to the clinical form of CVST are necessary for elucidating the pathophysiology of the disease. In this study, we aimed to establish a rat model that accurately recapitulates the clinical features of CVST in human patients. METHODS: This study consisted of a clinical analysis and animal experiments. Clinical data for two centres obtained between January 2016 and May 2021 were collected and analysed retrospectively. In addition, a Sprague-Dawley rat model of CVST was established by inserting a water-swellable rubber device into the superior sagittal sinus, following which imaging, histological, haematological, and behavioural tests were used to investigate pathophysiological changes. Principal component analysis and hierarchical clustering heatmaps were used to evaluate the similarity between the animal models and human patients. RESULTS: The imaging results revealed the possibility of vasogenic oedema in animal models. Haematological analysis indicated an inflammatory and hypercoagulable state. These findings were mostly matched with the retrospective clinical data. Pathological and serological tests further revealed brain parenchymal damage related to CVST in animal models. CONCLUSIONS: We successfully established a stable and reproducible rat model of CVST. The high similarity between clinical patients and animal models was verified via cluster analysis. This model may be useful for the study of CVST pathophysiology and potential therapies.

Defective 2D silicon phosphide monolayers for the nitrogen reduction reaction: a DFT study.

Electroreduction of N2 is a highly promising route for NH3 production. The lack of efficient catalysts that can activate and then reduce N2 into NH3 limits this as a pragmatic application. In this work, a 2D layered group IV-V material, silicon phosphide (SiP), is evaluated as a suitable substrate for the electrochemical nitrogen reduction reaction (ENRR). To capture N2, one phosphorus (P) defect was introduced on the plane of SiP. DFT calculations found that the defective SiP monolayer (D1-SiP, which is defined by the P-defect on SiP) exhibits enormous prospects towards the ENRR because of enhanced electron conductivity, good activation on N2, lower limiting potential (UL = -0.87 V) through the enzymatic pathway, smooth charge transfer between the catalyst and the reaction species, and robust thermal stability. Importantly, D1-SiP demonstrates the suppressed activities on producing of H2 and N2H4 side-products. This research demonstrates the potential of 2D metal-free Si-based catalysts for nitrogen fixation and further enriches the study of group IV-V materials for the ENRR.

A Hierarchical Spatial-Temporal Embedding Method Based on Enhanced Trajectory Features for Ship Type Classification.

Ship type classification is an essential task in maritime navigation domains, contributing to shipping monitoring, analysis, and forecasting. Presently, with the development of ship positioning and monitoring systems, many ship trajectory acquisitions make it possible to classify ships according to their movement pattern. Existing methods of ship classification based on trajectory include classical sequence analysis and deep learning methods. However, the real ship trajectories are unevenly distributed in geographical space, which leads to many problems in inferring the ship movement mode on the original ship trajectory. This paper proposes a hierarchical spatial-temporal embedding method based on enhanced trajectory features for ship type classification. We first preprocess the trajectory and combine the port information to transform the original ship trajectory into the moored records of ships, removing the unevenly distributed points in the trajectory data and enhancing key points' semantic information. Then, we propose a Hierarchical Spatial-Temporal Embedding Method (Hi-STEM) for ship classification. Hi-STEM maps moored records in the original geographical space into the feature space and can efficiently find the classification plane in the feature space. Experiments are conducted on real-world datasets and compared with several existing methods. The result shows that our approach has high accuracy in ship classification on ship moored records. We make the source code and datasets publicly available.

Genetic mutation analysis of 22 patients with congenital absence of vas deferens: a single-center study†.

Congenital absence of the vas deferens (CAVD), a congenital malformation of the male reproductive system, causes obstructive azoospermia and male infertility. Currently, the cystic fibrosis transmembrane conductance regulator (CFTR) has been recognized as the main pathogenic gene in CAVD, with some other genes, such as adhesion G-protein-coupled receptor G2 (ADGRG2), solute carrier family 9 isoform 3 (SLC9A3), sodium channel epithelial 1 subunit beta (SCNN1B), and carbonic anhydrase 12 (CA12), being candidate genes in the pathogenesis of CAVD. However, the frequency and spectrum of these mutations, as well as the pathogenic mechanisms of CAVD, have not been fully investigated. Here, we sequenced all genes with potentially pathogenic mutations using next-generation sequencing and verified all identified variants by Sanger sequencing. Further bioinformatic analysis was performed to predict the pathogenicity of mutations. We described the distribution of the p.V470M, poly-T, and TG-repeat CFTR polymorphisms and identified novel missense mutations in the CFTR and SLC9A3 genes, respectively. Taken together, we identified mutations in the CFTR, ADGRG2, SLC9A3, SCNN1B, and CA12 genes in 22 patients with CAVD, thus broadening the genetic spectrum of Chinese patients with CAVD.

The value of ultrasonography combined with carbohydrate antigen 125 and 19-9 detection in the diagnosis of borderline ovarian tumors and prediction of recurrence.

Objective: This study aimed to investigate the clinical value of ultrasonography combined with tumor markers in the diagnosis and prediction of recurrence of borderline ovarian tumors (BOTs) and analyze the value of the combination of two different auxiliary examinations in the diagnosis and prediction of recurrence of BOTs. Methods: Here, 221 patients with BOTs confirmed by postoperative pathology were enrolled. Their clinical data, including the ultrasonography features, tumor markers, and clinicopathological data, were retrospectively analyzed. Results: The statistical data of the 221 cases with BOTs were as follows: 94 (42.5%) with left-sided lesions, 102 (46.2%) with right-sided lesions, and 25 (11.3%) with bilateral lesions. Moreover, 93 cases (42.1%) had a borderline serous tumor, 110 (49.8%) had a borderline mucinous tumor, 12 (5.4%) had a borderline serous mucinous tumor, 2 (0.9%) had a borderline endometrioid tumor, 1 (0.5%) had a borderline Brenner tumor, and 2 (0.9%) had a clear cell BOT. There were 104 cases (47.1%) with a tumor diameter of ≤10 cm and 117 cases (52.9%) with a tumor diameter of >10 cm as suggested by ultrasonography. There were 89 cases (40.3%) with septation, 44 (19.9%) with papilla, and 97 (43.9%) with blood flow as demonstrated by ultrasonography. Carbohydrate antigen 125 (CA 125) was elevated in 132 cases (59.7%), and CA 19-9 was elevated in 52 cases (23.5%). Conclusion: In general, BOTs are difficult to diagnose preoperatively and have a certain recurrence rate. Ultrasonography combined with CA 125 and CA 19-9 is significant for the preoperative diagnosis and selection of surgical modality for BOTs and could be used as a guideline to achieve good preoperative preparation and avoid secondary surgery.

Artificial intelligence: A powerful paradigm for scientific research.

Artificial intelligence (AI) coupled with promising machine learning (ML) techniques well known from computer science is broadly affecting many aspects of various fields including science and technology, industry, and even our day-to-day life. The ML techniques have been developed to analyze high-throughput data with a view to obtaining useful insights, categorizing, predicting, and making evidence-based decisions in novel ways, which will promote the growth of novel applications and fuel the sustainable booming of AI. This paper undertakes a comprehensive survey on the development and application of AI in different aspects of fundamental sciences, including information science, mathematics, medical science, materials science, geoscience, life science, physics, and chemistry. The challenges that each discipline of science meets, and the potentials of AI techniques to handle these challenges, are discussed in detail. Moreover, we shed light on new research trends entailing the integration of AI into each scientific discipline. The aim of this paper is to provide a broad research guideline on fundamental sciences with potential infusion of AI, to help motivate researchers to deeply understand the state-of-the-art applications of AI-based fundamental sciences, and thereby to help promote the continuous development of these fundamental sciences.

p-Block element-doped silicon nanowires for nitrogen reduction reaction: a DFT study.

Photocatalytic nitrogen reduction reaction (NRR) is a promising, green route to chemically reducing N2 into NH3 under ambient conditions, correlating to the N2 fixation process of nitrogenase enzymes. To achieve high-yield NRR with sunlight as the driving force, high-performance photocatalysts are essential. One-dimensional silicon nanowires (1D SiNWs) are a great photoelectric candidate, but inactive for NRR due to their inability to capture N2. In this study, we proposed SiNWs doped by p-block elements (B, C, P) to tune the affinity to N2 and demonstrated that two-coordinated boron (B2C) offers an ultra-low overpotential (η) of 0.34 V to catalyze full NRR, which is even much lower than that of flat benchmark Ru(0001) catalysts (η = 0.92 V). Moreover, aspects including suppressed hydrogen evolution reaction (HER), high-spin ground state of the B2C site, and decreased band gap after B-doping ensure the high selectivity and photocatalytic activity. Finally, this work not only shows the potential use of metal-free p-block element-based catalysts, but also would facilitate the development of 1D nanomaterials towards efficient reduction of N2 into NH3.