Search

We present the results of the regression contest organized for the LIBS 2022 conference. While the motivation and design of the contest are briefly presented, the work focuses on the methodologies of the three best-performing teams. The employed spectral preprocessing strategies, choice of regression models and its optimization are detailed for each team separately. The aim of the contest reflects the long-term challenges faced by quantitative laser-induced breakdown spectroscopy (LIBS) analysis. Thus, the contest was designed with the purpose of providing a transparent platform for comparing and evaluating the large range of data processing tools available in the LIBS literature. Namely, the contest consisted of the quantification of two major (Cr, Ni) and two minor (Mn, Mo) elements in 15 steel targets. For constructing an appropriate regression model, spectra of 42 targets were provided. The spectra were collected using a commercially available laboratory-based LIBS system and made publicly available. The contest lasted 53 days during which the teams did not receive feedback. In total, 21 teams participated out of which the three best-performing methodologies are presented here. A single linear partial least squares model and two artificial neural network regression models are presented. The corresponding feature selection strategies included emission line selection, spectral range selection, and automatized wavelength selection. Various spectral normalization strategies and data augmentation strategies are also presented. JV acknowledges the financial support provided through the grant TACR TREND 6 - FW06010042 (Research and development of an advanced interaction vacuum system for laser spectroscopy). PP and EK acknowledge the financial support provided through the grant NCK II - TN02000020 (CAEPO: Center for advanced electronics and photon optics).

To date, various neural methods have been proposed for causal effect estimation based on observational data, where a default assumption is the same distribution and availability of variables at both training and inference (i.e., runtime) stages. However, distribution shift (i.e., domain shift) could happen during runtime, and bigger challenges arise from the impaired accessibility of variables. This is commonly caused by increasing privacy and ethical concerns, which can make arbitrary variables unavailable in the entire runtime data and imputation impractical. We term the co-occurrence of domain shift and inaccessible variables runtime domain corruption, which seriously impairs the generalizability of a trained counterfactual predictor. To counter runtime domain corruption, we subsume counterfactual prediction under the notion of domain adaptation. Specifically, we upper-bound the error w.r.t. the target domain (i.e., runtime covariates) by the sum of source domain error and inter-domain distribution distance. In addition, we build an adversarially unified variational causal effect model, named VEGAN, with a novel two-stage adversarial domain adaptation scheme to reduce the latent distribution disparity between treated and control groups first, and between training and runtime variables afterwards. We demonstrate that VEGAN outperforms other state-of-the-art baselines on individual-level treatment effect estimation in the presence of runtime domain corruption on benchmark datasets.

To address privacy concerns and reduce network latency, there has been a recent trend of compressing cumbersome recommendation models trained on the cloud and deploying compact recommender models to resource-limited devices for real-time recommendation. Existing solutions generally overlook device heterogeneity and user heterogeneity. They either require all devices to share the same compressed model or the devices with the same resource budget to share the same model. However, even users with the same devices may have different preferences. In addition, they assume the available resources (e.g., memory) for the recommender on a device are constant, which is not reflective of reality. In light of device and user heterogeneities as well as dynamic resource constraints, this paper proposes a Personalized Elastic Embedding Learning framework (PEEL) for on-device recommendation, which generates personalized embeddings for devices with various memory budgets in once-for-all manner, efficiently adapting to new or dynamic budgets, and effectively addressing user preference diversity by assigning personalized embeddings for different groups of users. Specifically, it pretrains using user-item interaction instances to generate the global embedding table and cluster users into groups. Then, it refines the embedding tables with local interaction instances within each group. Personalized elastic embedding is generated from the group-wise embedding blocks and their weights that indicate the contribution of each embedding block to the local recommendation performance. PEEL efficiently generates personalized elastic embeddings by selecting embedding blocks with the largest weights, making it adaptable to dynamic memory budgets. Extensive experiments are conducted on two public datasets, and the results show that PEEL yields superior performance on devices with heterogeneous and dynamic memory budgets.

Multi-modal knowledge graph (MKG) includes triplets that consist of entities and relations and multi-modal auxiliary data. In recent years, multi-hop multi-modal knowledge graph reasoning (MMKGR) based on reinforcement learning (RL) has received extensive attention because it addresses the intrinsic incompleteness of MKG in an interpretable manner. However, its performance is limited by empirically designed rewards and sparse relations. In addition, this method has been designed for the transductive setting where test entities have been seen during training, and it works poorly in the inductive setting where test entities do not appear in the training set. To overcome these issues, we propose TMR (Topology-aware Multi-hop Reasoning), which can conduct MKG reasoning under inductive and transductive settings. Specifically, TMR mainly consists of two components. (1) The topology-aware inductive representation captures information from the directed relations of unseen entities, and aggregates query-related topology features in an attentive manner to generate the fine-grained entity-independent features. (2) After completing multi-modal feature fusion, the relation-augment adaptive RL conducts multi-hop reasoning by eliminating manual rewards and dynamically adding actions. Finally, we construct new MKG datasets with different scales for inductive reasoning evaluation. Experimental results demonstrate that TMP outperforms state-of-the-art MKGR methods under both inductive and transductive settings.

Recent legislation of the "right to be forgotten" has led to the interest in machine unlearning, where the learned models are endowed with the function to forget information about specific training instances as if they have never existed in the training set. Previous work mainly focuses on computer vision scenarios and largely ignores the essentials of unlearning in NLP field, where text data contains more explicit and sensitive personal information than images. In this paper, we propose a general unlearning framework called KGA to induce forgetfulness. Different from previous work that tries to recover gradients or forces models to perform close to one specific distribution, KGA maintains distribution differences (i.e., knowledge gap). This relaxes the distribution assumption. Furthermore, we first apply the unlearning method to various NLP tasks (i.e., classification, translation, response generation) and propose several unlearning evaluation metrics with pertinence. Experiments on large-scale datasets show that KGA yields comprehensive improvements over baselines, where extensive analyses further validate the effectiveness of KGA and provide insight into unlearning for NLP tasks., Accepted to ACL 2023

Traditional recommender systems estimate user preference on items purely based on historical interaction records, thus failing to capture fine-grained yet dynamic user interests and letting users receive recommendation only passively. Recent conversational recommender systems (CRSs) tackle those limitations by enabling recommender systems to interact with the user to obtain her/his current preference through a sequence of clarifying questions. Despite the progress achieved in CRSs, existing solutions are far from satisfaction in the following two aspects: 1) current CRSs usually require each user to answer a quantity of clarifying questions before reaching the final recommendation, which harms the user experience; 2) there is a semantic gap between the learned representations of explicitly mentioned attributes and items. To address these drawbacks, we introduce the knowledge graph (KG) as the auxiliary information for comprehending and reasoning a user's preference, and propose a new CRS framework, namely Knowledge Enhanced Conversational Reasoning (KECR) system. As a user can reflect her/his preference via both attribute- and item-level expressions, KECR closes the semantic gap between two levels by embedding the structured knowledge in the KG. Meanwhile, KECR utilizes the connectivity within the KG to conduct explicit reasoning of the user demand, making the model less dependent on the user's feedback to clarifying questions. KECR can find a prominent reasoning chain to make the recommendation explainable and more rationale, as well as smoothen the conversation process, leading to better user experience and conversational recommendation accuracy. Extensive experiments on two real-world datasets demonstrate our approach's superiority over state-of-the-art baselines in both automatic evaluations and human judgments.

Temporal knowledge graphs (TKGs) model the temporal evolution of events and have recently attracted increasing attention. Since TKGs are intrinsically incomplete, it is necessary to reason out missing elements. Although existing TKG reasoning methods have the ability to predict missing future events, they fail to generate explicit reasoning paths and lack explainability. As reinforcement learning (RL) for multi-hop reasoning on traditional knowledge graphs starts showing superior explainability and performance in recent advances, it has opened up opportunities for exploring RL techniques on TKG reasoning. However, the performance of RL-based TKG reasoning methods is limited due to: (1) lack of ability to capture temporal evolution and semantic dependence jointly; (2) excessive reliance on manually designed rewards. To overcome these challenges, we propose an adaptive reinforcement learning model based on attention mechanism (DREAM) to predict missing elements in the future. Specifically, the model contains two components: (1) a multi-faceted attention representation learning method that captures semantic dependence and temporal evolution jointly; (2) an adaptive RL framework that conducts multi-hop reasoning by adaptively learning the reward functions. Experimental results demonstrate DREAM outperforms state-of-the-art models on public dataset, 11 pages

Latent factor models are the most popular backbones for today's recommender systems owing to their prominent performance. Latent factor models represent users and items as real-valued embedding vectors for pairwise similarity computation, and all embeddings are traditionally restricted to a uniform size that is relatively large (e.g., 256-dimensional). With the exponentially expanding user base and item catalog in contemporary e-commerce, this design is admittedly becoming memory-inefficient. To facilitate lightweight recommendation, reinforcement learning (RL) has recently opened up opportunities for identifying varying embedding sizes for different users/items. However, challenged by search efficiency and learning an optimal RL policy, existing RL-based methods are restricted to highly discrete, predefined embedding size choices. This leads to a largely overlooked potential of introducing finer granularity into embedding sizes to obtain better recommendation effectiveness under a given memory budget. In this paper, we propose continuous input embedding size search (CIESS), a novel RL-based method that operates on a continuous search space with arbitrary embedding sizes to choose from. In CIESS, we further present an innovative random walk-based exploration strategy to allow the RL policy to efficiently explore more candidate embedding sizes and converge to a better decision. CIESS is also model-agnostic and hence generalizable to a variety of latent factor RSs, whilst experiments on two real-world datasets have shown state-of-the-art performance of CIESS under different memory budgets when paired with three popular recommendation models., To appear in SIGIR'23

We introduce a new volatility model for option pricing that combines Markov switching with the realized generalized autoregressive conditional heteroskedasticity (GARCH) framework. This leads to a novel pricing kernel with a state-dependent variance risk premium and a pricing formula for European options, which is derived with an analytical approximation method. We apply the Markov-switching Realized GARCH model to Standard and Poor's 500 index options from 1990 to 2019 and find that investors' aversion to volatility-specific risk is time-varying. The proposed framework outperforms competing models and reduces (in-sample and out-of-sample) option-pricing errors by 15% or more.

Systematic review with network meta-analysis.We explored the efficacy and safety of different drug treatments in patients with spinal-cord injury (SCI)-related neuropathic pain. We investigated which treatment is most suitable for such patients by judging the efficacy and safety of these drugs.We searched the PubMed, Medline, Embase and Cochrane databases from inception to 31 August 2020. The quality of the included studies was assessed. We selected the proportion of patients whose pain was reduced by ≥50% and the prevalence of adverse effects as the outcome indicators of efficacy and safety, respectively.We included 15 randomized controlled clinical trials involving five interventions (anticonvulsants, antidepressants, anesthetics, opioids and botulinum toxin A). Based on the proportion of patients with pain reduction ≥50%, the order (from highest to lowest) was anticonvulsants anesthetics antidepressants botulinum toxin A opioidsplacebo. With regard to the prevalence of adverse effects, the order of safety (from highest to lowest) was placebo antidepressants botulinum toxin A anticonvulsants opioids anesthetics. Analyzes of efficacy and safety revealed that anticonvulsant, antidepressant and botulinum toxin A have good efficacy and safety.The efficacy of anticonvulsants, anesthetics, antidepressants, opioids and botulinum toxin A was greater than that of placebo for treatment of SCI-related neuropathic pain. However, the prevalence of adverse effects associated with use of these drugs was also higher than that of placebo. Further analyses based on efficacy and safety revealed anticonvulsants to be more suitable for such patients. In addition, antidepressant and botulinum toxin A may be promising treatments for SCI-related neuropathic pain, however, their effects still need to be further explored due to the small sample size.

Heterogeneous graph neural networks (HGNNs) have exhibited exceptional efficacy in modeling the complex heterogeneity in heterogeneous information networks (HINs). The critical advantage of HGNNs is their ability to handle diverse node and edge types in HINs by extracting and utilizing the abundant semantic information for effective representation learning. However, as a widespread phenomenon in many real-world scenarios, the class-imbalance distribution in HINs creates a performance bottleneck for existing HGNNs. Apart from the quantity imbalance of nodes, another more crucial and distinctive challenge in HINs is semantic imbalance. Minority classes in HINs often lack diverse and sufficient neighbor nodes, resulting in biased and incomplete semantic information. This semantic imbalance further compounds the difficulty of accurately classifying minority nodes, leading to the performance degradation of HGNNs. To tackle the imbalance of minority classes and supplement their inadequate semantics, we present the first method for the semantic imbalance problem in imbalanced HINs named Semantic-aware Node Synthesis (SNS). By assessing the influence on minority classes, SNS adaptively selects the heterogeneous neighbor nodes and augments the network with synthetic nodes while preserving the minority semantics. In addition, we introduce two regularization approaches for HGNNs that constrain the representation of synthetic nodes from both semantic and class perspectives to effectively suppress the potential noises from synthetic nodes, facilitating more expressive embeddings for classification. The comprehensive experimental study demonstrates that SNS consistently outperforms existing methods by a large margin in different benchmark datasets.

Stiff person syndrome is a rare autoimmune disease of the central nervous system that manifests as stiffness and painful spasms of the trunk axis and lower limb muscles. Benzodiazepines are the first choice for the clinical treatment of the disease. We reported a case of SPS. The patient presented with stiffness and convulsions of lower limbs, weakness after convulsions, falling off easily, abdominal muscle stiffness, and painful spasms lasting for several minutes and alleviating spontaneously. This was caused or aggravated by fatigue or mental stimulation. Left stiffness and weakness of the muscle after relief was present. A chest-enhanced computed tomography scan (CT) suggested two large ground glass nodules in the upper lobe of the left lung. Biopsy pathology indicated the nodules as adenocarcinoma in situ. The patient’s symptoms were significantly relieved after treatment with clonazepam and diazepam combined with pregabalin. The clinical manifestations of SPS vary among patients. The symptoms of the disease are mild or severe. Early identification and treatment can improve the prognoses of these patients.

Supplementary Material 7

Submission note: A thesis submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy to the School of Applied Systems Biology, La Trobe University, Victoria.

Supplementary Material 10

Supplementary Material 9

Supplementary Material 8

The analysis of iodine content in kelp is mainly focused on determining total iodine content, which lacks a rapid and sensitive analytical method for iodine species corresponding to iodate, iodide, monoiodotyrosine (MIT), and diiodo-tyrosine (DIT). Based on exploring the pretreatment methods of enzymatic hydrolysis of kelp, a rapid determination method of high-performance liquid chromatography and inductively coupled plasma mass spectrometry (HPLC-ICP-MS) for different iodine species existed in kelp samples was established. The pretreatment method is trypsin hydrolysis of kelp samples treated at 50°C for 10 hours, then the extraction solution was obtained by centrifugation, and filtration was proved for the first time. A high-efficiency separation anion-exchange column (Dionex IonPac AS-14) for iodate, iodide, MIT, and DIT with 0.1 mol·L−1(NH4)2CO3 solution at pH = 9 as a mobile phase, a flow rate of 0.8 mL·min−1 for the first 1500 s, and 1.2 mL·min−1 for the last 1800 s was confirmed. The linear ranges of the standard calibrations on four iodine species are 1.0–150 μg·L−1, and the detection limits are 1.62, 1.02, 1.59, and 2.55 μg·L−1, respectively, while the RSD values of four iodine species are all less than 2%. Under the optimized conditions, the proposed method was successfully applied for the speciation analysis of iodine in kelp samples.

Supplementary Material 11

Wireless capsule endoscopy (WCE) is a painless and non-invasive diagnostic tool for gastrointestinal (GI) diseases. However, due to GI anatomical constraints and hardware manufacturing limitations, WCE vision signals may suffer from insufficient illumination, leading to a complicated screening and examination procedure. Deep learning-based low-light image enhancement (LLIE) in the medical field gradually attracts researchers. Given the exuberant development of the denoising diffusion probabilistic model (DDPM) in computer vision, we introduce a WCE LLIE framework based on the multi-scale convolutional neural network (CNN) and reverse diffusion process. The multi-scale design allows models to preserve high-resolution representation and context information from low-resolution, while the curved wavelet attention (CWA) block is proposed for high-frequency and local feature learning. Furthermore, we combine the reverse diffusion procedure to further optimize the shallow output and generate the most realistic image. The proposed method is compared with ten state-of-the-art (SOTA) LLIE methods and significantly outperforms quantitatively and qualitatively. The superior performance on GI disease segmentation further demonstrates the clinical potential of our proposed model. Our code is publicly accessible., Comment: To appear in MICCAI 2023. Code availability: https://github.com/longbai1006/LLCaps

As an indispensable personalized service in Location-based Social Networks (LBSNs), the next Point-of-Interest (POI) recommendation aims to help people discover attractive and interesting places. Currently, most POI recommenders are based on the conventional centralized paradigm that heavily relies on the cloud to train the recommendation models with large volumes of collected users' sensitive check-in data. Although a few recent works have explored on-device frameworks for resilient and privacy-preserving POI recommendations, they invariably hold the assumption of model homogeneity for parameters/gradients aggregation and collaboration. However, users' mobile devices in the real world have various hardware configurations (e.g., compute resources), leading to heterogeneous on-device models with different architectures and sizes. In light of this, We propose a novel on-device POI recommendation framework, namely Model-Agnostic Collaborative learning for on-device POI recommendation (MAC), allowing users to customize their own model structures (e.g., dimension \& number of hidden layers). To counteract the sparsity of on-device user data, we propose to pre-select neighbors for collaboration based on physical distances, category-level preferences, and social networks. To assimilate knowledge from the above-selected neighbors in an efficient and secure way, we adopt the knowledge distillation framework with mutual information maximization. Instead of sharing sensitive models/gradients, clients in MAC only share their soft decisions on a preloaded reference dataset. To filter out low-quality neighbors, we propose two sampling strategies, performance-triggered sampling and similarity-based sampling, to speed up the training process and obtain optimal recommenders. In addition, we design two novel approaches to generate more effective reference datasets while protecting users' privacy.

Supplementary Material 12

Knowledge graphs (KGs) have become important auxiliary information for helping recommender systems obtain a good understanding of user preferences. Despite recent advances in KG-based recommender systems, existing methods are prone to suboptimal performance due to the following two drawbacks: 1) current KG-based methods over-emphasize the heterogeneous structural information within a KG and overlook the underlying semantics of its connections, hindering the recommender from distilling the explicit user preferences; and 2) the inherent incompleteness of a KG (i.e., missing facts, relations and entities) will deteriorate the information extracted from KG and weaken the representation learning of recommender systems. To tackle the aforementioned problems, we investigate the potential of jointly incorporating the structural and semantic information within a KG to model user preferences in finer granularity. A new framework for KG-based recommender systems, namely \textit{K}nowledge \textit{I}nfomax \textit{R}ecommender \textit{S}ystem with \textit{C}ontrastive \textit{L}earning (KIRS-CL) is proposed in this paper. Distinct from previous KG-based approaches, KIRS-CL utilizes structural and connectivity information with high-quality item embeddings learned by encoding KG triples with a pre-trained language model. These well-trained entity representations enable KIRS-CL to find the item to recommend via the preference connection between the user and the item. Additionally, to improve the generalizability of our framework, we introduce a contrastive warm-up learning strategy, making it capable of dealing with both warm- and cold-start recommendation scenarios. Extensive experiments on two real-world datasets demonstrate remarkable improvements over state-of-the-art baselines.

Objective To observe the effects of distance training on patients with continuous ambulatory peritoneal dialysis (CAPD) during the transmission control of the COVID-19 epidemic in China. Materials and methods CAPD patients from Xijing Hospital received a traditional training before the transmission control of COVID-19 epidemic, while they received a distance training dominated by WeChat and telephone during the transmission control of COVID-19 epidemic in China. Incidence and cure rate of PD-related peritonitis and catheter-related non-infectious complications were compared. All patients were followed up for 30 days from the date of complication. Results PD-related peritonitis, catheter displacement, and catheter occlusion had no significant difference, and the cure rate of PD-related peritonitis, catheter displacement, and catheter occlusion also had no significant difference in two comparisons despite the cure rate of PD-related peritonitis being slightly higher before (90.9%) than during (80%) the transmission control of COVID-19 epidemic. Conclusion Distance training mode had a similar effect compared to the traditional training mode in the prevention and treatment of PD-related peritonitis and catheter-related non-infectious complications. Practice implications Distance training model is an effective training mode that can be implemented in a short time during the epidemic period of serious infectious diseases.

One main theme in current condensed matter physics is the search of quantum spin liquid (QSL), an exotic magnetic state with strongly-fluctuating and highly-entangled spins down to zero temperature without static order. However, there is no consensus on the existence of a QSL ground state in any real material so far. The disorders and competing exchange interactions may prevent the formation of an ideal QSL state on frustrated spin lattices. Here we report systematic heat transport measurements on a honeycomb-lattice compound BaCo2(AsO4)2, which manifests magnetic order in zero field. In a narrow field range after the magnetic order is nearly suppressed by an in-plane field, in both perpendicular and parallel to the zigzag direction, a finite residual linear term of thermal conductivity is clearly observed, which is attributed to the mobile fractionalized spinon excitations. This provides smoking-gun evidence for a gapless QSL state in BaCo2(AsO4)2. We discuss the underlying physics to form this exotic gapless QSL state in Co2+ honeycomb lattice.

Objective: This network meta-analysis aimed to explore the effect of different drugs on mortality and neurological improvement in patients with traumatic brain injury (TBI), and to clarify which drug might be used as a more promising intervention for treating such patients by ranking.Methods: We conducted a comprehensive search from PubMed, Medline, Embase, and Cochrane Library databases from the establishment of the database to 31 January 2022. Data were extracted from the included studies, and the quality was assessed using the Cochrane risk-of-bias tool. The primary outcome measure was mortality in patients with TBI. The secondary outcome measures were the proportion of favorable outcomes and the occurrence of drug treatment–related side effects in patients with TBI in each drug treatment group. Statistical analyses were performed using Stata v16.0 and RevMan v5.3.0.Results: We included 30 randomized controlled trials that included 13 interventions (TXA, EPO, progesterone, progesterone + vitamin D, atorvastatin, beta-blocker therapy, Bradycor, Enoxaparin, Tracoprodi, dexanabinol, selenium, simvastatin, and placebo). The analysis revealed that these drugs significantly reduced mortality in patients with TBI and increased the proportion of patients with favorable outcomes after TBI compared with placebo. In terms of mortality after drug treatment, the order from the lowest to the highest was progesterone + vitamin D, beta-blocker therapy, EPO, simvastatin, Enoxaparin, Bradycor, Tracoprodi, selenium, atorvastatin, TXA, progesterone, dexanabinol, and placebo. In terms of the proportion of patients with favorable outcomes after drug treatment, the order from the highest to the lowest was as follows: Enoxaparin, progesterone + vitamin D, atorvastatin, simvastatin, Bradycor, EPO, beta-blocker therapy, progesterone, Tracoprodi, TXA, selenium, dexanabinol, and placebo. In addition, based on the classification of Glasgow Outcome Scale (GOS) scores after each drug treatment, this study also analyzed the three aspects of good recovery, moderate disability, and severe disability. It involved 10 interventions and revealed that compared with placebo treatment, a higher proportion of patients had a good recovery and moderate disability after treatment with progesterone + vitamin D, Bradycor, EPO, and progesterone. Meanwhile, the proportion of patients with a severe disability after treatment with progesterone + vitamin D and Bradycor was also low.Conclusion: The analysis of this study revealed that in patients with TBI, TXA, EPO, progesterone, progesterone + vitamin D, atorvastatin, beta-blocker therapy, Bradycor, Enoxaparin, Tracoprodi, dexanabinol, selenium, and simvastatin all reduced mortality and increased the proportion of patients with favorable outcomes in such patients compared with placebo. Among these, the progesterone + vitamin D had not only a higher proportion of patients with good recovery and moderate disability but also a lower proportion of patients with severe disability and mortality. However, whether this intervention can be used for clinical promotion still needs further exploration.

Introduction This study was performed to explore the efficacy and safety of pregabalin and gabapentin in patients with spinal cord injury (SCI)-induced neuropathic pain to determine which treatment is most suitable for such patients. Methods We searched the PubMed, MEDLINE, Embase, and Cochrane Library databases from database inception to August 31, 2020. The quality of the included studies was assessed. We selected the average pain intensity after treatment and the proportion of patients who discontinued treatment because of adverse effects as the outcome indicators for efficacy and safety, respectively. Statistical analyses were performed using Stata, v16.0, and RevMan, v5.3, software. Results We included eight randomized controlled trials that examined four interventions (pregabalin, gabapentin, carbamazepine, and amitriptyline). Based on the average pain intensity after treatment, the efficacy order from highest to lowest was pregabalin, gabapentin, amitriptyline, carbamazepine, and placebo. Based on the proportion of patients who discontinued treatment because of adverse effects, the order from highest to lowest was pregabalin, amitriptyline, carbamazepine, gabapentin, and placebo. In addition, five studies reported the overall incidence of treatment-related adverse effects for two interventions (pregabalin and gabapentin). According to the pooled analysis of these studies, the order for the overall incidence of treatment-related adverse effects from highest to lowest was pregabalin, gabapentin, and placebo. Conclusions This study revealed that for patients with SCI-related neuropathic pain, pregabalin was the most effective for relieving pain, whereas gabapentin performed better in aspects associated with drug therapy-related safety. Supplementary Information The online version contains supplementary material available at 10.1007/s40122-021-00302-8.

This paper is a study on a novel configuration switchable hydraulically interconnected suspension (CSHIS) system based on nonlinear model predictive control (NMPC). The proposed CSHIS system has si...

We show that the realized GARCH model yields closed-form expression for both the volatility index (VIX) and the volatility risk premium (VRP). The realized GARCH model is driven by two shocks, a return shock and a volatility shock, and these are natural state variables in the stochastic discount factor (SDF). The volatility shock endows the exponentially affine SDF with compensation for volatility risk. This leads to dissimilar dynamic properties under the physical and risk-neutral measures that can explain time-variation in the VRP. In an empirical application with the S&P 500 returns, the VIX, and the VRP, we find that the realized GARCH model significantly outperforms conventional GARCH models.

Background The significance of S100A8/A9 and S100A12 in anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) has not been clarified. This study was dedicated to exploring the potential pathogenic roles of S100A8/A9 and S100A12 in patients with myeloperoxidase (MPO)-ANCA-positive vasculitis. Methods Serum and urine concentrations of S100A8/A9 and S100A12 of forty-two AAV patients were evaluated. The influence of S100A8/A9 and S100A12 on the chemotaxis, the apoptosis, the release of IL-1β, the complement activation, the respiratory burst, as well as the neutrophil extracellular traps (NETs) formation of MPO-ANCA-activated neutrophils was investigated. Results The serum and urine S100A8/A9 and S100A12 of active MPO-AAV significantly increased (compared with inactive AAV and healthy controls, p Conclusion S100A8/A9 and S100A12 serve as markers for assessing the disease severity, and they may also play a role in MPO-AAV pathogenesis.

Contrastive learning (CL) has recently been demonstrated critical in improving recommendation performance. The underlying principle of CL-based recommendation models is to ensure the consistency between representations derived from different graph augmentations of the user-item bipartite graph. This self-supervised approach allows for the extraction of general features from raw data, thereby mitigating the issue of data sparsity. Despite the effectiveness of this paradigm, the factors contributing to its performance gains have yet to be fully understood. This paper provides novel insights into the impact of CL on recommendation. Our findings indicate that CL enables the model to learn more evenly distributed user and item representations, which alleviates the prevalent popularity bias and promoting long-tail items. Our analysis also suggests that the graph augmentations, previously considered essential, are relatively unreliable and of limited significance in CL-based recommendation. Based on these findings, we put forward an eXtremely Simple Graph Contrastive Learning method (XSimGCL) for recommendation, which discards the ineffective graph augmentations and instead employs a simple yet effective noise-based embedding augmentation to generate views for CL. A comprehensive experimental study on four large and highly sparse benchmark datasets demonstrates that, though the proposed method is extremely simple, it can smoothly adjust the uniformity of learned representations and outperforms its graph augmentation-based counterparts by a large margin in both recommendation accuracy and training efficiency. The code and used datasets are released at https://github.com/Coder-Yu/SELFRec., Extension of our SIGIR'22 paper "Are Graph Augmentations Necessary? Simple Graph Contrastive Learning for Recommendation", Accepted by TKDE

Summary: This project will examine the association between developmental trajectories of emotional problems during childhood and adolescence and early adult alcohol use problems, while controlling for developmental trajectories of conduct problems in childhood and adolescence. Data will be drawn from the Twins Early Development Study, a longitudinal twin study that follows twins born in England and Wales between 1994 and 1996, with assessments of emotional and conduct problems from 4-16 years old, and alcohol use problems at 21 years old. Latent growth curve models will be used to characterize developmental trajectories of emotional and conduct problems. These models will be embedded into a single structural equation model to incorporate alcohol use problems as the outcome. Multivariate biometric models will be used to decompose associations among trajectories of emotional problems, trajectories of conduct problems, and alcohol use problems at age 21 into genetic, shared and nonshared environmental influences to understand etiology of these associations. Background: A developmental pathway from emotional problems in childhood and adolescence to alcohol use problems in early adulthood was proposed by Hussong et al. (2011), but empirical evidence for this longitudinal association has been mixed; this may arise from an insufficient focus on comorbid externalizing problems (e.g., conduct problems) (Ning et al., 2020). Although the proposed project only includes conduct problems, we review the broader literature on externalizing problems because of its theoretical relevance. While studies that did not control for externalizing problems often reported positive associations between child/adolescent emotional problems and early adult problematic alcohol use (Berg et al., 2019; McKenzie et al., 2011; Pesola et al., 2014), studies that included externalizing problems as a covariate have reported negative or non-significant associations (Borges et al., 2018; Foster et al., 2018; Steele et al., 1995). For example, Foster et al. (2018) reported a negative direct effect from emotional problems to alcohol use problems, but emotional problems had a positive indirect association with alcohol use problems via increased externalizing problems. These findings suggest that the shared variance between emotional and externalizing problems may increase the risk for problematic alcohol use, while the unique variance of emotional problems may reduce the risk for problematic alcohol use. Among studies examining both emotional and externalizing problems, only a few have modelled developmental changes in these. Previous studies have suggested that emotional and externalizing problems both show changes in symptom levels across childhood and adolescence, and these changes vary by individuals (Hannigan et al., 2018; Nivard et al., 2017). Therefore, measuring emotional and externalizing problems at a single timepoint may not well-represent an individual’s symptom levels across the entire childhood and adolescence, which also obscures individual differences in these developmental changes. Studies modeling developmental changes of emotional and externalizing problems have reported that these changes in symptom levels across childhood and adolescence were associated with higher levels of alcohol use in early adulthood (Edwards et al., 2014; Timmermans et al., 2008). Gender differences have also been reported, such that the changes in depressive symptoms across adolescence were only positively associated with early adult harmful alcohol use in females but not males (Edwards et al., 2014). Among the few studies that have examined the association between developmental trajectories of emotional/externalizing problems and early adult alcohol use, most of them modelled developmental changes continuously, without considering potential developmental transitions that could occur from childhood to adolescence. There have been studies suggesting that a piecewise growth model, which allows different linear slopes in childhood and adolescence, fit better than a latent growth curve model assuming continuous changes in symptom levels across childhood and adolescence, for both emotional and externalizing problems (Cohen et al., 2018; Stepp et al., 2012). When predicting alcohol use problems, distinguishing developmental changes of emotional/externalizing problems in childhood from changes in adolescence could provide valuable information for timing of interventions. With this aim, the current study will use data from a longitudinal twin study (Twins Early Development Study; Rimfeld et al., 2019) to describe developmental changes in emotional and conduct problems from ages 4 to 16, using piecewise latent growth curve models to model changes in childhood (ages 4-9 years) and adolescence (ages 9-16 years) separately. We will then examine associations among intercepts (initial symptom levels) and slopes (changes of symptoms) of emotional and conduct problems during childhood and adolescence and alcohol use problems at age 21. Finally, we will explore gender differences in these associations. In addition to understanding how developmental trajectories of emotional and conduct problems may be longitudinally associated with early adult alcohol use problems, the current study seeks to understand why these behaviors may be associated in terms of genetic and/or environmental influences. Twin studies using biometric models in adult samples have consistently reported shared genetic factors among emotional problems, externalizing problems, and alcohol use problems (Kendler et al., 2003; Torvik et al., 2019; Torvik et al., 2017), although Kendler et al. (2003) found that nonshared environmental influences contributed more substantially to the covariance between emotional problems and alcohol use problems. A study that examined associations between externalizing problems in childhood/adolescence and alcohol use problems in adulthood also supported that shared genetic factors explained a large proportion of the covariance, with little covariance explained by environmental influences (Edwards & Kendler, 2012). On the other hand, Edwards et al. (2011) did not find substantial evidence for genetic, shared or nonshared environmental influences explaining the association between emotional problems in adolescence and alcohol intoxication frequency in early adulthood, probably due to the low magnitude of phenotypic correlations. None of these studies, however, has examined genetic and environmental associations between early adult alcohol use and developmental changes of emotional and externalizing problems. Given the dynamic nature of genetic and environmental influences on emotional and externalizing problems across childhood and adolescence (Hicks et al., 2007; Kendler et al., 2008), more genetically informed studies accounting for the development of these behavioral problems are needed. Therefore, the current study will use multivariate biometric models to decompose associations between alcohol use problems and developmental changes of emotional and conduct problems into genetic, shared and nonshared environmental factors. Understanding the mechanisms of associations among emotional problems, conduct problems and alcohol use problems would inform theory of change in prevention program designs.

The B-dot sensor is a type of Rogowski coil widely used in the measurement of current. However, the accuracy of the B-dot for measuring aircraft high-frequency lightning current is greatly affected by factors such as numerical integration drift, high-frequency oscillation, and calibration. In this study, a new design and optimization for improving the B-dot measuring accuracy was carried out. To correct the drift of the numerical integral of the measurement signal in differential mode, the measuring current was reconstructed based on the nonlinear least squares method. The sensor was then optimized by isolating the sampling resistance and matching the impedance with a voltage follower. A low-cost coaxial loop calibration system was also designed to calibrate the high frequency and strong magnetic fields more accurately. Finally, the optimized B-dot sensor accuracy was greatly improved with a measuring range of 30 kA/m, an error of 3.1%, and a high-frequency response of 50 MHz. Our study greatly increases the accuracy of measuring aircraft high-frequency lightning current.

With the rapid urbanization process, the construction of lighting facilities is increasing, whereas artificial light at nighttime (ALAN) negatively affects organisms in protected areas and threatens ecosystems. Therefore, a deep research of ALAN within protected areas is significant for better preserving biodiversity by scientific ALAN management. Taking the ecological conservation redline (ECR) in Zhejiang Province as a case study, we consistently applied remotely sensed ALAN data from 2000 to 2020 for exploring spatiotemporal changing characteristics of ALAN. More importantly, both human living and ecological safety were considered to classify ALAN status in 2019 in order to propose rational suggestions for management. The results showed ALAN intensified and expanded, increasing from 3.05 × 1012 nW·sr−1 to 5.24 × 1013 nW·sr−1 at an average growth rate of 2.35 × 1012 nW·sr−1·year−1. Hotspot analysis and bivariate spatial clustering identified the aggregation situation of ALAN and the population. They showed that statistically significant ALAN hotspots accounted for only 20.40% of the study area while providing 51.82% of the total ALAN. Based on the mismatches between human demand and ALAN supply, two crucial areas were identified where regulation is needed most, and targeted policy recommendations were put forward. The study results can contribute to the effective regulation of ALAN in protected areas.

Tunable Diode Laser Absorption Spectroscopy (TDLAS) has been widely applied in in situ and real-time monitoring of trace gas concentrations. In this paper, an advanced TDLAS-based optical gas sensing system with laser linewidth analysis and filtering/fitting algorithms is proposed and experimentally demonstrated. The linewidth of the laser pulse spectrum is innovatively considered and analyzed in the harmonic detection of the TDLAS model. The adaptive Variational Mode Decomposition-Savitzky Golay (VMD-SG) filtering algorithm is developed to process the raw data and could significantly eliminate the background noise variance by about 31% and signal jitters by about 12.5%. Furthermore, the Radial Basis Function (RBF) neural network is also incorporated and applied to improve the fitting accuracy of the gas sensor. Compared with traditional linear fitting or least squares method (LSM), the RBF neural network brings along the enhanced fitting accuracy within a large dynamic range, achieving an absolute error of below 50 ppmv (about 0.6%) for the maximum 8000 ppmv methane. The proposed technique in this paper is universal and compatible with TDLAS-based gas sensors without hardware modification, allowing direct improvement and optimization for current optical gas sensors.

Memory is an essential element for a computer to process information, which is integrated by logical circuits. Like electronic computing, thermal information can also be stored and read out by a thermal memory. Here, we show that a phase-changing polymer with hysteretic thermal transport properties can be experimentally processed into thermal memories at room temperature. We used a temperature-responsive and reversible polymer synthesized with melamine (M) and 6,7-dimethoxy-2,4[1H,3H]-quinazolinedione (Q) as a model system to demonstrate the manipulation of thermal transport at a molecular level. Fourier transform infrared spectroscopy and differential scanning calorimetry measurements indicate that this hysteretic behavior is based on the interaction of hydrogen bonds at high (317 K) and low (297 K) temperatures. This work demonstrates a controllable phonon transport process through the manipulation of hydrogen bonds, and thus it has potential applications in thermal memories.

Deep equilibrium models are based on implicitly defined functional relations and have shown competitive performance compared with the traditional deep networks. Monotone operator equilibrium networks (monDEQ) retain interesting performance with additional theoretical guaranties. Existing certification tools for classical deep networks cannot directly be applied to monDEQs for which much fewer tools exist. We introduce a semialgebraic representation for ReLU based monDEQs which allows to approximate the corresponding input output relation by semidefinite programming (SDP). We present several applications to network certification and obtain SDP models for the following problems : robustness certification, Lipschitz constant estimation, ellipsoidal uncertainty propagation. We use these models to certify robustness of monDEQs w.r.t. a general $L_q$ norm. Experimental results show that the proposed models outperform existing approaches for monDEQ certification. Furthermore, our investigations suggest that monDEQs are much more robust to $L_2$ perturbations than $L_{\infty}$ perturbations., 16 pages, 4 tables, 2 figures

Purpose: To explore the efficacy of nimodipine, nifedipine, and edaravone (EDA) combined with Nao-Xue-Shu in patients with hypertensive intracerebral hemorrhage (HICH) and to determine the best western medicine combined with Nao-Xue-Shu for treating HICH patients using a ranking method.Methods: After a comprehensive search of the China National Knowledge Infrastructure (CNKI), Wanfang Database, VIP information database, Chinese Biomedical Database (CBM), PubMed, Embase, and Cochrane Library database from the database establishment 31 December 2021, data extraction and quality assessment were conducted for the included articles. The primary outcome measure was the effectiveness after treatment. Secondary outcome measures were after-treatment the National Institutes of Health Stroke Scale (NIHSS) scores, hematoma volume, perihematoma edema volume, and inflammatory factor expression levels. Statistical analyses were performed using Stata 16.0 and RevMan 5.3.0 software.Results: We included 19 randomized controlled trials (RCTs) and six non-RCTs. The effective rate after treatment was ranked from the best to the worst as follows: routine cure measure (RCM) + nifedipine + Nao-Xue-Shu, RCM + EDA + Nao-Xue-Shu, RCM + Nao-Xue-Shu, RCM + nimodipine + Nao-Xue-Shu, RCM + EDA, and RCM. The post-treatment NHISS scores from lowest to highest were as follows: RCM + EDA + Nao-Xue-Shu, RCM + nifedipine + Nao-Xue-Shu, RCM + EDA, RCM + nimodipine + Nao-Xue-Shu, RCM + Nao-Xue-Shu, RCM + Nao-Xue-Kang, and RCM. The post-treatment hematoma volume from minimum to maximum was as follows: RCM + EDA + Nao-Xue-Shu, RCM + nimodipine + Nao-Xue-Shu, RCM + nifedipine + Nao-Xue-Shu, RCM + Nao-Xue-Shu, RCM + Nao-Xue-Kang, and RCM. The post-treatment perihematoma edema volume from minimum to maximum was as follows: RCM + EDA + Nao-Xue-Shu, RCM + nifedipine + Nao-Xue-Shu, RCM + nimodipine + Nao-Xue-Shu, RCM + Nao-Xue-Shu, and RCM. For inflammatory factor expression levels after treatment, IL-6 concentration levels after treatment from lowest to highest wasas follows: RCM + Nao-Xue-Shu, RCM + nifedipine + Nao-Xue-Shu, RCM + nimodipine + Nao-Xue-Shu, RCM + EDA + Nao-Xue-Shu, and RCM. TNF-α concentration levels after treatment from lowest to highest was as follow: RCM + nimodipine + Nao-Xue-Shu, RCM + nifedipine + Nao-Xue-Shu, RCM + Nao-Xue-Shu, and RCM.Conclusion: Nao-Xue-Shu combined with nifedipine showed better effectiveness after treatment in HICH patients compared with the other combinations. Nao-Xue-Shu combined with EDA was more effective for improving neurological function and reducing both hematoma and edema volumes around the hematoma compared with the other combinations. However, Nao-Xue-Shu alone or Nao-Xue-Shu combined with nimodipine may be more effective for reducing proinflammatory factor expression.

Surgical methods for treating spasmodic torticollis include denervation and myotomy (DAM) of the affected muscles, and deep brain stimulation (DBS). This study reports on the long-term efficacy, prognostic factors, safety, and hospitalization costs following these two procedures. We collected data from 94 patients with spasmodic torticollis, of whom 41 and 53 were treated with DAM and DBS, respectively, from June 2008 to December 2020 at the Chinese People’s Liberation Army General Hospital. We used the Tsui scale and the global outcome score of the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) to evaluate the preoperative and postoperative clinical condition in all patients. We also determined the cost of hospitalization, prognostic factors, and serious adverse events following the two surgical procedures. The mean follow-up time was 68.83 months (range = 13–116). Both resection surgery and DBS showed good results in terms of Tsui (Z = -5.103, P = 0.000; Z = -6.210, P = 0.000) and TWSTRS scores (t = 8.762, P = 0.000; Z = -6.308, P = 0.000). Compared with the DAM group, the preoperative (47.71, range 24–67.25) and postoperative (18.57, range 0–53) TWSTRS scores in the DBS group were significantly higher (Z = -3.161, P = 0.002). We found no correlation between prognostic factors and the patient age, gender, or disease duration for either surgical procedure. However, prognostic factors were related to the length of the postoperative follow-up period in the DBS surgery group (Z = -2.068, P = 0.039; Z = -3.287, P = 0.001). The mean hospitalization cost in the DBS group was 6.85 times that found in the resection group (Z = -8.284, P = 0.000). The total complication rate was 4.26%. We found, both resection surgery and DBS, showed good results in the patients with spasmodic torticollis. Although DBS had a greater impact on the quality of life and the degree of pain relief, it was more expensive. Prognostic factors were related to the length of the postoperative follow-up period in patients who underwent DBS surgery.

BackgroundThis research aimed to build an m6A-associated lncRNA prognostic model of esophageal cancer that can be used to predict outcome in esophageal cancer patients.MethodsRNA sequencing transcriptome data and clinical information about patients with esophageal cancer were obtained according to TCGA. Twenty-four m6A-associated genes were selected based on previous studies. m6A-associated lncRNAs were determined through Pearson correlation analysis. Three m6A-associated lncRNA prognostic signatures were built through analysis of the training set using univariate, LASSO, and multivariate Cox regression. To validate the stabilization of the risk signature, Kaplan–Meier and ROC curve analyses were performed on the testing and complete sets. The prognoses of EC patients were predicted quantitatively by building a nomogram. GSEA was conducted to analyze the underlying signaling pathways and biological processes. To identify the underlying mechanisms through which the lncRNAs act, we constructed a PPI network and a ceRNA network and conducted GO and KEGG pathway analyses. EC samples were evaluated using the ESTIMATE algorithm to compute stromal, immune, and estimate scores. The ssGSEA algorithm was used to quantitatively infer immune cell infiltration and immune functions. The TIDE algorithm was performed to simulate immune evasion and predict the response to immunotherapy.ResultsWe identified and validated an m6A-associated lncRNA risk model in EC that could correctly and reliably predict the OS of EC patients. The ceRNA network, PPI network, and GO and KEGG pathway analyses confirmed and the underlying mechanisms and functions provided enlightenment regarding therapeutic strategies for EC. Immunotherapy responses were better in the low-risk subgroup, and PD-1 and CTLA4 checkpoint immunotherapy benefited the patients in the low-risk subgroup.ConclusionsWe constructed a new m6A-related lncRNA prognostic risk model of EC, based on three m6A-related lncRNAs: LINC01612, AC025166.1 and AC016876.2, that can predict the prognoses of EC patients.

It is well known that realized measures of volatility, which are computed from high-frequency intraday data, provide accurate measurements of the latent volatility process. We investigate the role of realized volatility in pricing VIX options using the GARV model (Christoffersen et al., 2014), and the Realized GARCH model (Hansen et al., 2012). A closed-form pricing formula for the (affine) GARV model is developed. For the (non-affine) log-linear Realized GARCH model, we introduce a novel approximation approach to derive its analytical pricing formula. We show that the newly proposed approximation method is fast, with a high degree of accuracy. An extensive empirical application on VIX options from 2006 to 2020 shows that models with realized volatility significantly outperform conventional GARCH-type models based on daily returns only. Among these, the Realized GARCH model provides the best pricing performance due to its fewer constraints and a more flexible modeling structure. Our results hold both in-sample and out-of-sample.

In recent years, neural architecture-based recommender systems have achieved tremendous success, but they still fall short of expectation when dealing with highly sparse data. Self-supervised learning (SSL), as an emerging technique for learning from unlabeled data, has attracted considerable attention as a potential solution to this issue. This survey paper presents a systematic and timely review of research efforts on self-supervised recommendation (SSR). Specifically, we propose an exclusive definition of SSR, on top of which we develop a comprehensive taxonomy to divide existing SSR methods into four categories: contrastive, generative, predictive, and hybrid. For each category, we elucidate its concept and formulation, the involved methods, as well as its pros and cons. Furthermore, to facilitate empirical comparison, we release an open-source library SELFRec (https://github.com/Coder-Yu/SELFRec), which incorporates a wide range of SSR models and benchmark datasets. Through rigorous experiments using this library, we derive and report some significant findings regarding the selection of self-supervised signals for enhancing recommendation. Finally, we shed light on the limitations in the current research and outline the future research directions., 20 pages. Accepted by TKDE

A common problem in health research is that we have a large database with many variables measured on a large number of individuals. We are interested in measuring additional variables on a subsample; these measurements may be newly available, or expensive, or simply not considered when the data were first collected. The intended use for the new measurements is to fit a regression model generalisable to the whole cohort (and to its source population). This is a two-phase sampling problem; it differs from some other two-phase sampling problems in the richness of the phase I data and in the goal of regression modelling. In particular, an important special case is measurement-error models, where a variable strongly correlated with the phase II measurements is available at phase I. We will explain how influence functions have been useful as a unifying concept for extending classical results to this setting, and describe the steps from designing for a simple weighted estimator at known parameter values through adaptive multiwave designs and the use of prior information. We will conclude with some comments on the information gap between design-based and model-based estimators in this setting.