Search

Purpose: To evaluate the effectiveness of combined remifentanil/sevoflurane anesthesia on hemodynamic stability and postoperative recovery in paediatric tonsil surgery. Methods: In this prospective study, a total of 102 paediatric patients who underwent tonsil surgery in Taihe County People's Hospital between February 2021 and April 2022 were recruited and randomized at a ratio of 1:1 to receive either propofol alone (control group) or remifentanil plus sevoflurane (study group) during paediatric tonsil surgery, with 51 patients in each group. The anaesthetic effect of the two regimens was the primary endpoint and was evaluated by monitoring the mean arterial pressure (MAP), heart rate (HR) and oxygen saturation (SpO2) levels of the patients before anaesthesia (T0), immediately after intubation (T1), at the beginning of surgery (T2) and at extubation (T3). Results: Remifentanil plus sevoflurane resulted in more stable MAP and HR of patients at T1, T2 and T3 than propofol alone (p < 0.05). Furthermore, patients in the two groups exhibited similar SpO2 levels at T0, T1, T2 and T3 (p > 0.05). Conclusion: The combined use of remifentanil and sevoflurane anaesthetics is effective in paediatric patients undergoing tonsil surgery and thus may offer a viable alternative for such clinical practice. The drug combination maintains intraoperative haemodynamic stability in patients, boosts postoperative recovery, and lowers the risk of postoperative adverse events.

Complementary Labels Learning (CLL) arises in many real-world tasks such as private questions classification and online learning, which aims to alleviate the annotation cost compared with standard supervised learning. Unfortunately, most previous CLL algorithms were in a stable environment rather than an open and dynamic scenarios, where data collected from unseen augmented classes in the training process might emerge in the testing phase. In this paper, we propose a novel problem setting called Complementary Labels Learning with Augmented Classes (CLLAC), which brings the challenge that classifiers trained by complementary labels should not only be able to classify the instances from observed classes accurately, but also recognize the instance from the Augmented Classes in the testing phase. Specifically, by using unlabeled data, we propose an unbiased estimator of classification risk for CLLAC, which is guaranteed to be provably consistent. Moreover, we provide generalization error bound for proposed method which shows that the optimal parametric convergence rate is achieved for estimation error. Finally, the experimental results on several benchmark datasets verify the effectiveness of the proposed method.

Efficient and effective attacks are crucial for reliable evaluation of defenses, and also for developing robust models. Projected Gradient Descent (PGD) attack has been demonstrated to be one of the most successful adversarial attacks. However, the effect of the standard PGD attack can be easily weakened by rescaling the logits, while the original decision of every input will not be changed. To mitigate this issue, in this paper, we propose Scale-Invariant Adversarial Attack (SI-PGD), which utilizes the angle between the features in the penultimate layer and the weights in the softmax layer to guide the generation of adversaries. The cosine angle matrix is used to learn angularly discriminative representation and will not be changed with the rescaling of logits, thus making SI-PGD attack to be stable and effective. We evaluate our attack against multiple defenses and show improved performance when compared with existing attacks. Further, we propose Scale-Invariant (SI) adversarial defense mechanism based on the cosine angle matrix, which can be embedded into the popular adversarial defenses. The experimental results show the defense method with our SI mechanism achieves state-of-the-art performance among multi-step and single-step defenses., Comment: TDSC under review

BACKGROUND: Coronary microcirculation has a fundamental role in the regulation of coronary blood flow in response to cardiac requirements, which has aroused wide concerns in basic science and clinical cardiovascular research. We aimed to analyze coronary microcirculation-associated literatures over 30 years and provide insightful information on the evolutionary path, frontier research hotspots, and future developmental trends. METHODS: Publications were retrieved from the Web of Science Core Collection (WoSCC). VOSviewer was used to perform co-occurrence analyses for countries, institutions, authors, and keywords and to generate visualized collaboration maps. CiteSpace was used to visualize the knowledge map derived from reference co-citation analysis, burst references, and keywords detection. RESULTS: This analysis was performed based on 11,702 publications including 9981 articles and 1721 reviews. The United States and Harvard University ranked at the top among all the countries and institutions. The majority of articles were published in Circulation, and it also was the most co-cited journal. Thematic hotspots and frontiers were focused on coronary microvascular dysfunction, magnetic resonance imaging, fractional flow reserve, STEMI, and heart failure. Additionally, keywords burst and co-occurrence cluster analysis showed that management, microvascular dysfunction, microvascular obstruction, prognostic value, outcomes, and guidelines were current knowledge gaps and future directions. CONCLUSIONS: Coronary microcirculation presented a research hotspot relevant wide spectrum of cardiovascular diseases. Definite diagnostics and prognostics are particularly valued. The protection of cardiovascular events that influence clinical outcomes should be an insightful concern in the future. Multidisciplinary collaborations will provide significant advances for the development of coronary microcirculation.

Many studies have addressed the protection of source location privacy (SLP). However, most of the traditional research has not achieved a balance between security and energy consumption. Here, we propose a dynamic ring-based routing (DRBR) scheme to solve this problem. This scheme is divided into three stages. In the first stage, the source node randomly selects an intermediate node to send a data packet to the selected area. During the second stage, the intermediate node sends the data packet to the mixing ring, where the ring nodes combine data packets. The mixed data packets from the mixing ring are sent to the sink node in the third stage. We prove that DRBR not only provides high SLP, but also achieves a desirable trade-off between privacy and energy consumption.

Objective(s): The current study investigated the potential hepatoprotective effects of the ethanolic extracts of Chinese propolis (EECP) on ethanol-induced fatty liver in mice.Materials and Methods: C57BL/6J mice were orally gavaged with 50% ethanol alone or co-administrated with EECP at the dose of 0.2 ml/kg bodyweight for eight weeks. The dose for ethanol was 6 ml/kg bodyweight for the first two experimental weeks, and then increased to 8, 10, and 12 ml/kg bodyweight every two experimental weeks. Alterations in the hepatic transcriptome due to concomitant administration of EECP were investigated using RNA-Seq technique.Results: Our results showed that the main EECP-responsive genes were involved in lipid syntheses, which were significantly down-regulated in both female and male mice co-administrated with EECP. In female mice, these differentially expressed genes (DEGs) were mainly associated with fatty acid biosynthesis. While in male mice, these DEGs were mainly involved in the steroid metabolic process and cholesterol biosynthetic process. Despite the sex-associated responses in lipid metabolism, EECP also exerted other beneficial effects in female mice through modulation of the cytokine-cytokine receptor interaction pathway that helped explaining its hepato-protective effectiveness.Conclusion: Our findings indicated that the mechanism regarding the hepato-protective effects of EECP was gender-dependent, which is worthy of further investigation during the development of therapeutic interventions using EECP to reduce the adverse influences of ethanol.

Additional file 10.

Additional file 8.

Additional file 7.

Additional file 4.

Additional file 6.

Additional file 5.

Additional file 9.

A range of defense methods have been proposed to improve the robustness of neural networks on adversarial examples, among which provable defense methods have been demonstrated to be effective to train neural networks that are certifiably robust to the attacker. However, most of these provable defense methods treat all examples equally during training process, which ignore the inconsistent constraint of certified robustness between correctly classified (natural) and misclassified examples. In this paper, we explore this inconsistency caused by misclassified examples and add a novel consistency regularization term to make better use of the misclassified examples. Specifically, we identified that the certified robustness of network can be significantly improved if the constraint of certified robustness on misclassified examples and correctly classified examples is consistent. Motivated by this discovery, we design a new defense regularization term called Misclassification Aware Adversarial Regularization (MAAR), which constrains the output probability distributions of all examples in the certified region of the misclassified example. Experimental results show that our proposed MAAR achieves the best certified robustness and comparable accuracy on CIFAR-10 and MNIST datasets in comparison with several state-of-the-art methods., 4 pages, appeare in AAAI21-RSEML

With the rapid development of natural language and the implementation of the Wisdom Court, intelligent judicial assistants has become a new application of natural language processing in the judicial field. The text classification method based on word vector and deep neural network implements statistical classification of judicial documents, but it can not achieve the inherent logical interpretation of judicial cases. A method of extracting semantic logic tree from judicial case texts is proposed, and event tree can be interpreted by deep forest. Judicial documents are divided into several sub-tree fragments by sentence segmentation, and each sub-tree fragment is analyzed by dependency syntax to obtain core subject-predicate-object triples. TF-IDF algorithm is used to calculate the weights of triples, and get the core sub-event sequence, and use pruning algorithm to construct the max heap. The designed triple encoding algorithm realizes max heap vectorization of event tree, and embedded deep forest algorithm to realize classification discrimination of judicial text event tree. The experimental results show that the proposed event tree construction method combined with the deep forest algorithm can greatly improve the logical interpretation and accuracy of judicial text.

With the rapid development of artificial intelligence and its wide application on the ground, more and more devices is embedded in intelligent question answering technology. The development of mobile Internet technology and the popularization of community platform provide us with huge and jumbled data. Credibility has become an urgent problem to be solved in question answering system. As far as the medical platform is concerned, there is a huge noise problem in the current general medical information retrieval. Based on this, we propose a trusted evaluation mechanism based on the information of the medical community platform. In order to search and select more accurate and credible answers, we propose an optimization method of search ranking based on credible evaluation mechanism. By analyzing the professional competence of information providers, this method trustfully ranks multiple answers in the question-and-answer set of medical communities, which largely solves the interference of “one-question multi-answer” phenomenon in the question-and-answer system on retrieval. At the same time, it uses department classification, subdivides medical directions, and further matches the areas of concern of information providers, so as to improve the accuracy and hit rate of the retrieval system.

Since its introduction in 2003, the influence maximization (IM) problem has drawn significant research attention in the literature. The aim of IM is to select a set of k users who can influence the most individuals in the social network. The problem is proven to be NP-hard. A large number of approximate algorithms have been proposed to address this problem. The state-of-the-art algorithms estimate the expected influence of nodes based on sampled diffusion paths. As the number of required samples have been recently proven to be lower bounded by a particular threshold that presets tradeoff between the accuracy and efficiency, the result quality of these traditional solutions is hard to be further improved without sacrificing efficiency. In this paper, we present an orthogonal and novel paradigm to address the IM problem by leveraging deep learning models to estimate the expected influence. Specifically, we present a novel framework called DISCO that incorporates network embedding and deep reinforcement learning techniques to address this problem. Experimental study on real-world networks demonstrates that DISCO achieves the best performance w.r.t efficiency and influence spread quality compared to state-of-the-art classical solutions. Besides, we also show that the learning model exhibits good generality., 14 pages, 5 figures

Objectifs: Examiner les tendances temporelles, les déterminants en lien avec le design des études et la qualité des taux de réponse rapportés dans des études cas-témoins sur le cancer publiées lors des 30 dernières années. Méthodes: Une revue des études cas-témoins sur le cancer a été menée. Les critères d'inclusion étaient la publication (i) dans l’un de 15 grands périodiques ciblés et (ii) lors de quatre périodes de publication (1984-1986, 1995, 2005 et 2013) couvrant trois décennies. 370 études ont été sélectionnées et examinées. La méthodologie en lien avec le recrutement des sujets et la collecte de données, les caractéristiques de la population, les taux de participation et les raisons de la non-participation ont été extraites de ces études. Des statistiques descriptives ont été utilisées pour résumer la qualité des taux de réponse rapportés (en fonction de la quantité d’information disponible), les tendances temporelles et les déterminants des taux de réponse; des modèles de régression linéaire ont été utilisés pour analyser les tendances temporelles et les déterminants des taux de participation. Résultats: Dans l'ensemble, les qualités des taux de réponse rapportés et des raisons de non-participation étaient très faible, particulièrement chez les témoins. La participation a diminué au cours des 30 dernières années, et cette baisse est plus marquée dans les études menées après 2000. Lorsque l'on compare les taux de réponse dans les études récentes a ceux des études menées au cours de 1971 à 1980, il y a une plus grande baisse chez les témoins sélectionnés en population générale ( -17,04%, IC 95%: -23,17%, -10,91%) que chez les cas (-5,99%, IC 95%: -11,50%, -0,48%). Les déterminants statistiquement significatifs du taux de réponse chez les cas étaient: le type de cancer examiné, la localisation géographique de la population de l'étude, et le mode de collecte des données. Le seul déterminant statistiquement significatif du taux de réponse chez les témoins hospitaliers était leur localisation géographique. Le seul déterminant statistiquement significatif du taux de participation chez les témoins sélectionnés en population générale était le type de répondant (sujet uniquement ou accompagné d’une tierce personne). Conclusion: Le taux de participation dans les études cas-témoins sur le cancer semble avoir diminué au cours des 30 dernières années et cette baisse serait plus marquée dans les études récentes. Afin d'évaluer le niveau réel de non-participation et ses déterminants, ainsi que l'impact de la non-participation sur la validité des études, il est nécessaire que les études publiées utilisent une approche normalisée pour calculer leurs taux de participation et qu’elles rapportent ceux-ci de façon transparente., Objectives: To examine the time trends, study design determinants, and quality of reporting of response rates in published case-control studies of cancer over the past 30 years. Methods: A review was conducted of case-control studies of cancer. Inclusion criteria required publications in 15 major journals, during four publication periods spanning three decades (1984-86, 1995, 2005 and 2013). 370 studies were selected and reviewed. Information on study base ascertainment, data collection methods, population characteristics, response rates, and reasons for non-participation was extracted. Quality of response rate reporting was assessed based on the amount of information reported. Descriptive statistics were used to summarize the quality of the reporting, time trends and the determinants of response rates; linear regression models were used to analyse time trends and determinants of response rates. Results: Overall, the quality of reporting of response rates and reasons for non-participation was very poor, especially for control series. Participation has declined over the past 30 years, and this decline was steeper in studies conducted after 2000. When comparing the response rates in recent studies to that in studies conducted during 1971-1980, there was a greater decline of this rate in population controls (-17.04%, 95% CI: -23.17%, -10.91%) than in cases (-5.99%, 95% CI: -11.50%, -0.48%). Statistically significant determinants of response rates among cases were: cancer type examined, location of the study population, and mode of data collection. The only statistically significant determinant of response rates among medical source controls was location of the study population. The only statistically significant determinant of response rates among population controls was type of respondent (self only or self and proxy) accepted by studies. Conclusion: Response rates in case-control studies of cancer seem to have declined and this decline has accelerated in recent studies. In order to appreciate the true level of non-participation and its determinants, as well as the impact of non-participation on validity of studies, there is a need for more transparent reporting and standardized calculation of response rates in published studies.

Over more than a decade, IS research has examined the role of trust in the context of technology adoptions, such as website acceptance (Gefen et al. 2003), successful online interactions (Coppola et al. 2004) and recommendation agent usage (Wang and Benbasat 2005). The primary antecedent of trust in this line of research is trustee’s trustworthiness. However, Dashiti et al. (2011) articulated the effect of trust- received (trustor’s felt trust from trustee) on trust-given (trustor’s trust in trustee) and consequently the trusting behaviors in the e-government context. This study aims to investigate the context of online social networks (OSNs) and to test whether trust-received plays a similar role in forming the interpersonal trust among Facebook user’s interactions. Based on this trust reciprocity theory, this study aims to find out (i) whether users are more willing to trust others and share their personal information with them when individuals experienced trust-received and (ii) what factors, especially what IT features, influence trust-received in personal networking context. We hypothesize that information sensitivity and audience limitation are two main influencing factors over users’ trust-received on Facebook. We analyzed our hypotheses using an online scenario-based survey, and based on our findings we proposed a new IT artifact in the follow-up study to explore more trust-received influencing factors. We found that audience limitation (an IT artifact) has the most significant impact over trust-received and that reciprocity of trust does exist on Facebook. We also concluded that the mentioning of a virtual function in which the re-share could be switched on or off has an effect over trust-received under certain circumstances. Thus we concluded that IT artifacts can help build or maintain online users relationships and that trust can form from a reciprocity pattern.