Search

This report provides an introduction to the Computational Intelligence and Forecasting Technologies (CIFT) project at Lawrence Berkeley National Laboratory (LBNL). The main objective of CIFT is to promote the use of high-performance computing (HPC) tools and techniques for analysis of streaming data. After noticing the data volume being given as the explanation for the five-month delay for SEC and CFTC to issue their report on the 2010 Flash Crash, LBNL started the CIFT project to apply HPC technologies to manage and analyze financial data. Making timely decisions with streaming data is a requirement for many different applications, such as avoiding impending failure in the electric power grid or a liquidity crisis in financial markets. In all these cases, the HPC tools are well suited in handling the complex data dependencies and providing timely solutions. Over the years, CIFT has worked on a number of different forms of streaming data, including those from vehicle traffic, electric power grid, electricity usage, and so on. The following sections explain the key features of HPC systems, introduce a few special tools used on these systems, and provide examples of streaming data analyses using these HPC tools.

Automated network intrusion detection systems (NIDS) continuously monitor the network traffic to detect attacks or/and anomalies. These systems need to be able to detect attacks and alert network engineers in real-time. Therefore, modern NIDS are built using complex machine learning algorithms that require large training datasets and are time-consuming to train. The proposed work shows that machine learning algorithms from the RAPIDS cuML library on Graphics Processing Units (GPUs) can speed-up the training process on large scale datasets. This approach is able to reduce the training time while providing high accuracy and performance. We demonstrate the proposed approach on a large subset of data extracted from the Aegean Wi-Fi Intrusion Dataset (AWID). Multiple classification experiments were performed on both CPU and GPU. We achieve up to 65x acceleration of training several machine learning methods by moving most of the pipeline computations to the GPU and leveraging the new cuML library as well as the GPU version of the CatBoost library.

Large scientific projects are increasing relying on analyses of data for their new discoveries; and a number of different data management systems have been developed to serve this scientific projects. In the work-in-progress paper, we describe an effort on understanding the data access patterns of one of these data management systems, dCache. This particular deployment of dCache acts as a disk cache in front of a large tape storage system primarily containing high-energy physics data. Based on the 15-month dCache logs, the cache is only accessing the tape system once for over 50 file requests, which indicates that it is effective as a disk cache. The on-disk files are repeated used, more than three times a day. We have also identified a number of unusual access patterns that are worth further investigation.

The volume of data moving through a network increases with new scientific experiments and simulations. Network bandwidth requirements also increase proportionally to deliver data within a certain time frame. We observe that a significant portion of the popular dataset is transferred multiple times to different users as well as to the same user for various reasons. In-network data caching for the shared data has shown to reduce the redundant data transfers and consequently save network traffic volume. In addition, overall application performance is expected to improve with in-network caching because access to the locally cached data results in lower latency. This paper shows how much data was shared over the study period, how much network traffic volume was consequently saved, and how much the temporary in-network caching increased the scientific application performance. It also analyzes data access patterns in applications and the impacts of caching nodes on the regional data repository. From the results, we observed that the network bandwidth demand was reduced by nearly a factor of 3 over the study period.

The widely-adopted practice is to train deep learning models with specialized hardware accelerators, e.g., GPUs or TPUs, due to their superior performance on linear algebra operations. However, this strategy does not employ effectively the extensive CPU and memory resources - which are used only for preprocessing, data transfer, and scheduling - available by default on the accelerated servers. In this paper, we study training algorithms for deep learning on heterogeneous CPU+GPU architectures. Our two-fold objective - maximize convergence rate and resource utilization simultaneously - makes the problem challenging. In order to allow for a principled exploration of the design space, we first introduce a generic deep learning framework that exploits the difference in computational power and memory hierarchy between CPU and GPU through asynchronous message passing. Based on insights gained through experimentation with the framework, we design two heterogeneous asynchronous stochastic gradient descent (SGD) algorithms. The first algorithm - CPU+GPU Hogbatch - combines small batches on CPU with large batches on GPU in order to maximize the utilization of both resources. However, this generates an unbalanced model update distribution which hinders the statistical convergence. The second algorithm - Adaptive Hogbatch - assigns batches with continuously evolving size based on the relative speed of CPU and GPU. This balances the model updates ratio at the expense of a customizable decrease in utilization. We show that the implementation of these algorithms in the proposed CPU+GPU framework achieves both faster convergence and higher resource utilization than TensorFlow on several real datasets.

Atmospheric ammonia (NH3) can affect nitrogen deposition, particle acidity, and gas-particle partitioning. Although the inorganic chemistry of NH3 in fine particulate (PM2.5) formation are well-constrained, the understanding of interactions between NH3 and secondary organic aerosol (SOA) are rather insufficient until recently. Laboratory studies indicate that NH3 molecule can react with SOA then forms nitrogen-containing organic compounds (NOCs), which can further react to form heterocyclic organic compounds. In this study, we use a modified version of the CMAQ model to simulate the potential importance of the SOA-ammonia uptake mechanism on air quality over China in summer and winter 2017, considering a range of assumed NH3 uptake coefficients (10−3–10−5). Our results show that uptake of NH3 by SOA leads to a decrease in gas-phase NH3 mixing ratio, by as much as 27.5% and 19.0% for the highest uptake coefficient scenario (10−3) in summer and winter, respectively. The largest reduction of ammonia occurs over the Sichuan Basin and the North China Plain. The reduction of gas-phase NH3 engenders a decrease of ammonium nitrate, by up to 30%, but has little impact on the ammonium sulfate concentration. Uptake of NH3 does not significantly affect SOA concentrations owing to overall moderate changes in aerosol acidity, and thus small effects on SOA formation from isoprene. Altogether, NH3 uptake led to a reduction in the average PM2.5 concentration up to 8.9% and 8.7% for the highest uptake coefficient (10−3) in summer and winter, respectively. These results highlight the need for better constraints on the NH3-SOA interactions.

To meet the exascale I/O requirements for the High-Performance Computing (HPC), a new I/O subsystem, Burst Buffer, based on solid state drives (SSD), has been developed. However, the diverse HPC workloads and the bursty I/O pattern cause severe data fragmentation that requires costly garbage collection (GC) and increases the number of bytes written to the SSD. To address this data fragmentation challenge, a new multi-stream feature has been developed for SSDs. In this work, we develop an I/O Separation scheme called BIOS to leverage this multi-stream feature to group the I/O streams based on the user IDs. We propose a stream-aware scheduling policy based on burst buffer pools in the workload manager, and integrate the BIOS with the workload manager to optimize the I/O separation scheme in burst buffer. We evaluate the proposed framework with a burst buffer I/O traces from Cori Supercomputer including a diverse set of applications. Experimental results show that the BIOS could improve the performance by 1.44x on average and reduce the Write Amplification Factor (WAF) by up to 1.20x. These demonstrate the potential benefits of the I/O separation scheme for solid state storage systems.

Donghao Wang,1,* Yingying Ren,2,* Riken Chen,1,* Xiangxia Zeng,1,* Qiming Gan,1,* Zhiyang Zhuang,1 Xiaofen Su,1 Kang Wu,1 Sun Zhang,1 Yongkang Tang,1 Shiwei Li,1 Haojie Zhang,1,3 Yanyan Zhou,1 Nuofu Zhang,1 Dongxing Zhao1 1State Key Laboratory of Respiratory Disease, Sleep Medicine Center, Guangzhou Institute of Respiratory Health, National Center for Respiratory Disease, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, People’s Republic of China; 2Medical Records and Statistics Room, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, People’s Republic of China; 3The Clinical Medicine Department, Henan University, Zhengzhou, People’s Republic of China*These authors contributed equally to this workCorrespondence: Dongxing Zhao; Nuofu Zhang, State Key Laboratory of Respiratory Disease, National Clinical Research Center for Respiratory Disease, Sleep Medicine Center, Guangzhou Institute of Respiratory Health, National Center for Respiratory Disease, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, People’s Republic of China, Tel +86-13650901411 ; +86-13600460056, Email rieast@msn.com; nfzhanggird@163.comObjective: Obstructive sleep apnea (OSA) is a common sleep-disordered breathing disease. We aimed to establish an improved screening questionnaire without physical examinations for OSA named the CNCQ-OSA (Chinese community questionnaire for OSA).Methods: A total of 2585 participants who visited sleep medicine center and underwent overnight polysomnography were grouped into two independent cohorts: derivation (n = 2180) and validation (n = 405). The CNCQ-OSA was designed according to the baseline of patients in derivation cohort. We comprehensively analyzed the data to evaluate the predictive value of the CNCQ-OSA, compared to the GOAL questionnaire, STOP-Bang questionnaire (SBQ) and NoSAS questionnaire.Results: The CNCQ-OSA included seven variables: loud snoring, BMI ≥ 25 kg/m2, male gender, apnea, sleepiness, hypertension and age ≥ 30, with a total score ranging from 7 to 16.7 points (≥ 13.5 points indicating high risk of OSA, ≥ 14.5 points indicating extremely high risk). In the derivation and validation cohorts, the areas under the curve of the CNCQ-OSA were 0.761 and 0.767, respectively. In the validation cohort, the sensitivity and specificity of a CNCQ-OSA score ≥ 13.5 points for the apnea–hypopnea index (AHI) ≥ 5/h were 0.821 and 0.559, respectively (Youden index, 0.380), and the score ≥ 14.5 points were 0.494 and 0.887, respectively (Youden index, 0.375). The CNCQ-OSA had a better predictive value for AHI ≥ 5/h, AHI > 15/h and AHI > 30/h, with the highest Youden index, compared to the other questionnaires.Conclusion: The CNCQ-OSA can effectively identify the risk of OSA, which is appropriate for self-screening at home without physical examinations.Keywords: obstructive sleep apnea, screening, GOAL, STOP-Bang, NoSAS

Meijing Shen,* Xianghao Chen,* Jingyue He, Lin Xiong, Rengui Tian, Guangwu Yang, He Zha, Kaifeng Wu Department of Laboratory Medicine, the First People’s Hospital of Zunyi (The Third Affiliated Hospital of Zunyi Medical University), Zunyi, People’s Republic of China*These authors contributed equally to this workCorrespondence: Kaifeng Wu; He Zha, Department of Laboratory Medicine, the First People’s Hospital of Zunyi (The Third Affiliated Hospital of Zunyi Medical University), Zunyi, People’s Republic of China, Email kiphoonwu@126.com; zhahe666@126.comPurpose: Carbapenem-resistant Klebsiella pneumoniae (CRKP) has seriously threatened public health worldwide. This study aimed to investigate the antimicrobial resistance patterns, sequence types (STs), virulence and carbapenemase genes of CRKP isolates from patients in Zunyi, China.Methods: CRKP isolates were collected from the First People’s Hospital of Zunyi between January 2018 and December 2020. Antimicrobial susceptibility was determined using a VITEK® 2 analyzer and confirmed using either the broth dilution method, Kirby–Bauer method, or E-test assays. Carbapenemase production was examined using a modified carbapenem inactivation method. STs of the studied isolates were determined by multilocus sequence typing, and the presence of carbapenemase and virulence genes was examined using polymerase chain reaction assays.Results: In total, 94 CRKP isolates were collected. All studied isolates produced carbapenemase, and the most common carbapenemase gene was New Delhi metallo-β-lactamase (NDM; 72.3%), followed by Klebsiella pneumoniae carbapenemase (KPC; 24.5%), and Verona integron-encoded metallo-β-lactamase (VIM; 3.2%). Of the studied isolates, 74.3% exhibited multidrug-resistant (MDR) phenotype, and 25.7% were either pandrug-resistant (PDR) or extensively drug-resistant (XDR) phenotypes. The most prevalent sequence type was ST2407 (37.2%), followed by ST76 (21.3%) and ST11 (11.7%). The NDM gene was present in 97.1% of ST2407 isolates and 90.0% of ST76 isolates, whereas the KPC gene was present in 90.9% of ST11 isolates. The majority of the isolates carried wabG, uge, and fimH virulence genes, with prevalence rates of 94.7%, 92.6%, and 94.7%, respectively.Conclusion: This study describes NDM-producing ST2407 and ST76, as well as KPC-producing ST11, as the major clonal types of CRKP isolates in Zunyi, China. All CRKP isolates were resistant to multiple types of antibiotics, and the majority of isolates carried carbapenemase and virulence genes. Clonal spread of NDM-producing CRKP ST2407 and ST76, and KPC-producing CRKP ST11 should be strictly monitored.Keywords: Klebsiella pneumoniae, antimicrobial resistance patterns, virulence genes, sequence types, carbapenemase genes

Yang Liu,1 Chao Zhang,2 Qianwen Wang,1 Kangze Wu,3 Zhouyi Sun,1 Zhe Tang,1 Bo Zhang3 1The Fourth Affiliated Hospital, Zhejiang University School of Medicine, Yiwu, Zhejiang, 322000, People’s Republic of China; 2Center for Evidence-Based Medicine and Clinical Research, Taihe Hospital, Hubei University of Medicine, Shiyan, Hubei, 442000, People’s Republic of China; 3The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, Zhejiang, 310000, People’s Republic of ChinaCorrespondence: Bo Zhang, The Second Affiliated Hospital, Zhejiang University School of Medicine, 88 Jiefang Road, Hangzhou, Zhejiang, 310000, People’s Republic of China, Tel/Fax +86-0571-87783563, Email jjs10@zju.edu.cnBackground: This study aimed to evaluate the global colorectal cancer(CRC) trend and the relevant risk factors from 1990 to 2019 and for better policymaking and resource allocation.Methods: Data on CRC, including incidence, mortality and disability adjusted life year (DALY) rates, were extracted from the 2019 Global Burden of Disease (GBD) study. The estimated annual percentage changes (EAPCs) were calculated to assess the temporal trend of incidence, mortality and DALYs. The Bayesian age-period-cohort model(BAPC) was used to predict the future burden of CRC.Results: In 2019, a total of 2.17 million CRC cases were reported worldwide, a 157% increase from 1990. In high-social demographic index (SDI) regions, the trend of age-standardized incidence rate(ASIR) tended to decrease, while the proportion of people under 50 years of age tended to increase. Although the number of deaths and DALYs increased, the age-standardized death rate (ASDR) and age-standardized DALY rate decreased. The CRC burden was growing fastest in middle-SDI regions, especially in East Asia, followed by low SDI regions. In addition, the milk intake, High-BMI and high fasting plasma glucose play a more important role in on CRC. The predicted cases and deaths in global continued to increase to 2044. And there is an upward trend in ASIR for both men and women.Conclusion: In developed regions, the CRC burden continues to decrease, while the CRC burden become more and more severe in developing regions. Overall, the burden of CRC will rising in the near future. Therefore, reasonable resource allocation and prevention policies should be implemented. Developing countries needs more attention.Keywords: global burden of disease, colorectal cancer, disability-adjusted life year, estimated annual percentage change, age-standardized incidence rate, Bayesian age-period-cohort

While federated learning (FL) has gained great attention for mobile and Internet of Things (IoT) computing with the benefits of scalable cooperative learning and privacy protection capabilities, there still exist a great deal of technical challenges to make it practically deployable. For instance, the distribution of the training process to a myriad of devices limits the classification performance of machine learning (ML) algorithms, often showing a significantly degraded accuracy compared to centralized learning. In this paper, we investigate the problem of performance limitation under FL and present the benefit of data augmentation with an application of anomaly detection using an IoT dataset. Our initial study reveals that one of the critical reasons for the performance degradation is that each device sees only a small fraction of data (that it generates), which limits the efficacy of the local ML model (constructed by the device). This becomes more critical if the data holds the class imbalance problem, observed not infrequently in practice (e.g., a small fraction of anomalies). Moreover, device heterogeneity with respect to data quantity is an open challenge in FL. Based on these observations, we examine the impact of data augmentation on detection performance in FL settings (both homogeneous and heterogeneous). Our experimental results show that even a simple random oversampling can improve detection performance with manageable learning complexity.

To understand behaviors of natural and man-made events, such as energy consumption of buildings, which accounts for 40% of energy uses in the US, we deploy automated monitoring devices to record periodic observations. However, such experimental and observation data often contains problems and irregularities that have to be cleaned up before analyses. Due to various conditions affecting sensor operations, the communication channels, recording steps, or the recording media, the recorded data might have missing values, errors, or anomalous values. An effective way to clean up these problems is to replace these missing values, errors and anomalous values with expected values, a process generally known as imputation. In this work, we survey commonly used missing value imputation techniques and compare their performance on a set of building monitoring data. To compare the different types of sensor measurements with widely varying characteristics, we use normalized root mean squared error (NRMSE) as the key metric for the effectiveness of the imputation methods. We additionally consider periodicity and run time when considering comparing methods. Through extensive testing, we find that for small gap sizes, up to 8 consecutive missing values, linear interpolation performs the best; for larger gaps stretching up to 48 consecutive missing values, K-nearest neighbors provides the most accurate imputations; for even larger gaps, more computational intensive methods, such as matrix factorization, achieve the smallest NRMSE. Additionally, we observe that these computationally intensive algorithms not only provide accurate imputations for large gaps, but are also more robust across all types of sensors.

Botnet detection is an active research topic as botnets are a source of many malicious activities, including distributed denial-of-service (DDoS), click-fraud, spamming, and crypto-mining attacks. However, it is getting more complicated to identify botnets due to the continuous evolution of botnet software and families that harness new types of devices and attack vectors. Recent studies employing machine learning (ML) showed improved performance to detect botnets to some extent, but they are still limited and ineffective with the lack of sequential pattern analysis, which is a key to detect various classes of botnets. In this paper, we propose a novel botnet detection method, built upon Recurrent Variational Autoencoder (RVAE), that effectively captures sequential characteristics of botnet anomalies. We validate the feasibility of the proposed method with the CTU-13 dataset that have been widely employed for botnet detection studies, and show that our method is at least comparable to existing techniques in terms of detection accuracy. In addition, our experimental results show that the proposed method can detect previously unseen botnets by utilizing sequential patterns of network traffic. We will also show how our method can detect botnets in the streaming mode, which is the essential requirement to perform real-time, on-line detection.

Streamflow components (e.g., young water vs old water) and hydrological transit times play an important role in water resource and water quality management. We collected a four-year record of stable isotopes of oxygen and hydrogen (δ18O and δ2H) in precipitation, groundwater and stream water for six catchments in the Yongan watershed of eastern China. The stable isotope records were used to identify spatio-temporal variations in the young water fraction (Fyw, defined as the proportion of the transit-time distribution younger than a threshold age) and mean transit time (MTT) based on sine-wave fitting and convolution integral methods, respectively. The Fyw ranged from 14 to 35% in the Yongan watershed. Cumulative transit time showed contrasting distributions, suggesting considerable heterogeneity and a complex interplay between catchment characteristics. Estimated MTTs ranged from 3.2 to 6.3 years and may be explained by catchment characteristics (e.g., elevation and topographic gradient). Observed spatial trends in MTTs likely result from contrasting contributions of different-aged subsurface water flows across the six catchments. The use of Fyw constraints in estimating MTTs reduced uncertainty in some catchments, suggesting the potential benefits of combining multiple approaches (e.g., Fyw) to optimize the results of traditional calibration methods. The relatively low Fyw and long MTTs highlight the importance of groundwater contributions to streamflow generation and imply a considerable lag time in river water quantity and quality responses to catchment-scale water resource management. Coupling multiple metrics (e.g., Fyw and MTT) and isotope models enhances our understanding of watershed-scale hydrologic processes and hydrograph separation.

We present ADIOS 2, the latest version of the Adaptable Input Output (I/O) System. ADIOS 2 addresses scientific data management needs ranging from scalable I/O in supercomputers, to data analysis in personal computer and cloud systems. Version 2 introduces a unified application programming interface (API) that enables seamless data movement through files, wide-area-networks, and direct memory access, as well as high-level APIs for data analysis. The internal architecture provides a set of reusable and extendable components for managing data presentation and transport mechanisms for new applications. ADIOS 2 bindings are available in C++11, C, Fortran, Python, and Matlab and are currently used across different scientific communities. ADIOS 2 provides a communal framework to tackle data management challenges as we approach the exascale era of supercomputing.

Machine Learning (ML) methods have been widely used in Intrusion Detection Systems (IDS). In particular, many botnet detection methods are based on ML. However, due to the fast-evolving nature of network security threats, it is necessary to frequently retrain the ML tools with up-to-date data, especially because data labeling takes a long time and requires a lot of effort, making it difficult to generate training data. We propose transfer learning as a more effective approach for botnet detection, as it can learn from well curated source data and transfer the knowledge to a target problem domain not seen before. We devise an approach that is effective regardless whether or not the data from the target domain is labeled. More specifically, we train a neural network with the Recurrrent Variation Autoencoder (RVAE) structure on the source data, and use RVAE to compute anomaly scores for data records from the target domain. In an evaluation of this transfer learning framework, we use CTU-13 dataset as a source domain and a fresh set of network monitoring data as a target domain. Tests show that the proposed transfer learning method is able to detect botnets better than semi-supervised learning method that was trained on the target domain data. The area under Receiver Operating Characteristic is 0.810 for transfer learning, and 0.779 for directly using RVAE on the target domain data.

With the growth of 5G wireless technologies and IoT, it become urgent to develop robust network security systems, such as intrusions detection systems (IDS) to keep the networks secure. These IDS systems need to detect unauthorized access and attacks in real-time. However, most of the modern IDS are built based on complex machine learning models that are time-consuming to train. In this work, we propose a methodology using the SHapley Additive exPlanations (SHAP) in combination with tree-based classifiers. SHAP can be used to select consistent and small feature subsets to reduce the execution time and improve classification accuracy. We demonstrate the proposed approach with the Aegean Wi-Fi Intrusion Dataset (AWID) dataset in a series of multi-class classification experiments. Among the four classes ("normal", "injection", "flooding"and "impersonation"), it is well-known that the class impersonation is hard to be classified accurately. Tests show that we can use about 10% of the initial feature set without reducing the overall prediction accuracy. With this reduced set of features, the training time could be reduced as much as a factor of four, while slightly improving the discriminating ability to identify impersonation instances. This study suggests that by reducing the number of features, the classification algorithms are able to focus on key trends that differentiates the "attacks"classes from the "normal"class. Using a reduces subset of features improves IDS's accuracy and performance. Also, SHAP dependence plots capture the relationship between individual features and the classification decision.

Large-scale high performance computing (HPC) systems typically consist of many thousands of CPUs and storage units, while used by hundreds to thousands of users at the same time. Applications from these large numbers of users have diverse characteristics, such as varying compute, communication, memory, and I/O intensiveness. A good understanding of the performance characteristics of each user application is important for job scheduling and resource provisioning. Among these performance characteristics, the I/O performance is difficult to predict because the I/O system software is complex, the I/O system is shared among all users, and the I/O operations also heavily rely on networking systems. To improve the prediction of the I/O performance on HPC systems, we propose to integrate information from a number of different system logs and develop a regression-based approach that dynamically selects the most relevant features from the most recent log entries, and automatically select the best regression algorithm for the prediction task. Evaluation results show that our proposed scheme can predict the I/O performance with up to 84% prediction accuracy in the case of the I/O-intensive applications using the logs from CORI supercomputer at NERSC.

Large high-performance computers (HPC) are expensive tools responsible for supporting thousands of scientific applications. However, it is not easy to determine the best set of configurations for workloads to best utilize the storage and I/O systems. Users typically use the default configurations provided by the system administrators, which typically results in poor performance. In an effort to identify application characteristics more important to I/O performance, we applied several machine learning techniques to characterize these applications. To identify the features that are most relevant to the I/O performance, we evaluate a number of different feature selection methods, e.g., Mutual information regression and F regression, and develop a novel feature selection method based on Min-max mutual information. These feature selection methods allow us to sift through a large set of the real-world workloads collected from NERSC's Cori supercomputer system, and identify the most important features. We employ a number of different clustering algorithms, including KMeans, Gaussian Mixture Model (GMM) and Ward linkage, and measure the cluster quality with Davies Boulder Index (DBI), Silhouette and a new Combined Score developed for this work. The cluster evaluation result shows that the test dataset could be best divided into three clusters, where cluster 1 contains mostly small jobs with operations on standard I/O units, cluster 2 consists of middle size parallel jobs dominated by read operations, and cluster 3 include large parallel jobs with heavy write operations. The cluster characteristics suggest that using parallel I/O library MPI IO and a large number of parallel cores are important to achieve high I/O throughput.

We consider data sources connected to a software defined network (SDN) with heterogeneous link access rates. Deadline-driven data transfer requests are made to a centralized network controller that schedules pacing rates of sources and meeting the request deadline has a pre-assigned value. The goal of the scheduler is to maximize the aggregate value. We design a scheduler (RL-Agent) based on Deep Deterministic Policy Gradient (DDPG). We compare our approach with three heuristics: (i) PFAIR, which shares the bottleneck capacity in proportion to the access rates, (ii) VDRatio, which prioritizes flows with high value-to-demand ratio, and (iii) VBEDF, which prioritizes flows with high value-to-deadline ratio. For equally valued requests and homogeneous access rates, PFAIR is the same as an idealized TCP algorithm, while VBEDF and VDRatio reduce to the Earliest Deadline First (EDF) and the Shortest Job First (SJF) algorithms, respectively. In this scenario, we show that RL-Agent performs significantly better than PFAIR and VDRatio and matches and in over-loaded scenarios out-performs VBEDF. When access rates are heterogeneous, we show that the RL-Agent performs as well as VBEDF even though the RL-Agent has no knowledge of the heterogeneity to start with. For the value maximization problems, we show that the RL-Agent out-performs the heuristics for both homogeneous and heterogeneous access networks. For the general case of heterogeneity with different values, the RL-Agent performs the best despite having no prior knowledge of the heterogeneity and the values, whereas the heuristics have full knowledge of the heterogeneity and VDRatio and VBEDF have partial knowledge of the values through the ratios of value to demand and value to deadline, respectively.

Quickly identifying astronomical transients from synoptic surveys is critical to many recent astrophysical discoveries. However, each of the data processing pipelines in these surveys contains dozens of stages with highly varying time and space requirements. Properly predicting the resources required to run these pipelines is critical for the allocation of computing resources and reducing the discovery response time. We propose a machine learning strategy for this prediction task and demonstrate its effectiveness using a set of timing measurements from the intermediate Palomar Transient Factory (iPTF) workflow. The proposed model utilizes the spatiotemporal correlation of astronomical images, where nearby patches of the sky (space) are likely to have a similar number of objects of interest and workflows executed in the recent past (time) are likely to use a similar amount of time because the machines and data storage systems are likely to be in similar states. We capture the relationship among these spatial and temporal features in a Bayesian network and study how they impact the prediction accuracy. This Bayesian network helps us to identify the most influential features for predictions. With proper features, our models achieve errors close to the random variance boundary within batches of images taken at the same time, which can be regarded as the intrinsic limit of prediction accuracy.

Daily transportation mode choice is largely habitual, but transitions between life events may disrupt travel habits and can shift choices between alternative transportation modes. Although much is known about general mode switches following life event transitions, less is understood about differences that may exist between subpopulations, especially from a long-term perspective. Understanding these differences will help planners and policymakers introduce more targeted policy interventions to promote sustainable transportation modes and inform longer-term predictions. Extending beyond existing literature, we use data collected from a retrospective survey to investigate the effects of life course events on mode use situated within different long-term life trajectory contexts. We apply a machine-learning method called joint social sequence clustering to define five distinct and interpretable cohorts based on trajectory patterns in family and career domains over their life courses. We use these patterns as an innovative contextual system to investigate (1) the heterogeneous effects of life events on travel mode use and (2) further differentiation between gender and generation groups in these life event effects. We find that events occurring relatively early in life are more strongly associated with changes in mode-use behavior, and that mode use can also be affected by the relative order of events. This timing and order effect can have lasting impacts on mode use aggregated over entire life cycles: members of our ‘‘Have-it-alls’’ cohort—who finish their education, start working, partner up, and have children early in life—ramp up car use at each event, resulting in the highest rate of car use occurring the earliest among all the cohorts. Women drive more when having children primarily when their family formation and career formation are intertwined early in life, and younger generations rely relatively more on car use during familial events when their careers have a later start.

Today, scientific experiments and simulations produce massive amounts of heterogeneous data that need to be stored and analyzed. Given that these large datasets are stored in many files, formats and locations, how can scientists find relevant data, duplicates or similarities? In this context, we concentrate on developing algorithms to compare similarity of time series for the purpose of search, classification and clustering. For example, generating accurate patterns from climate related time series is important not only for building models for weather forecasting and climate prediction, but also for modeling and predicting the cycle of carbon, water, and energy. We developed the methodology and ran an exploratory analysis of climatic and ecosystem variables from the FLUXNET2015 dataset. The proposed combination of similarity metrics, nonlinear dimension reduction, clustering methods and validity measures for time series data has never been applied to unlabeled datasets before, and provides a process that can be easily extended to other scientific time series data. The dimensionality reduction step provides a good way to identify the optimum number of clusters, detect outliers and assign initial labels to the time series data. We evaluated multiple similarity metrics, in terms of the internal cluster validity for driver as well as response variables. While the best metric often depends on a number of factor, the Euclidean distance seems to perform well for most variables and also in terms of computational expense.

The demands of increasingly large scientific application workflows lead to the need for more powerful supercomputers. As the scale of supercomputing systems have grown, the prediction of fault tolerance has become an increasingly critical area of study, since the prediction of system failures can improve performance by saving checkpoints in advance. We propose a real-time failure detection algorithm that adopts an event-based prediction model. The prediction model is a convolutional neural network that utilizes both traditional event attributes and additional spatio-temporal features. We present a case study using our proposed method with six years of reliability, availability, and serviceability event logs recorded by Mira, a Blue Gene/Q supercomputer at Argonne National Laboratory. In the case study, we have shown that our failure prediction model is not limited to predict the occurrence of failures in general. It is capable of accurately detecting specific types of critical failures such as coolant and power problems within reasonable lead time ranges. Our case study shows that the proposed method can achieve a F1 score of 0.56 for general failures, 0.97 for coolant failures, and 0.86 for power failures.

Prediction of daily transportation mode use (car, public transit, or active travel) is a important task in transportation research. Unlike statistical models that impose a predetermined model structure, machine learning models are learned from the data, making them more flexible with higher prediction accuracy. However, prediction of mid-to long-term habitual modes still largely relies on traditional statistical analysis using small samples of cross-sectional data. Low interpretability of 'black-box' machine learning models limits their usefulness for generating behavior insights needed for designing appropriate interventions. This paper, leveraging a set of unique longitudinal life course data, is the first use case to demonstrate machine learning methods applied for both predicting and interpreting regularly used travel modes. We combine sequence clustering and tree-based machine learning methods coupled with TreeExplainer to predict and interpret habitual travel modes using mid-to long-term predictors. Five life course clusters are derived to provide evaluation and interpretation contexts. This allows us to improve upon a recently developed TreeExplainer method to better distinguish predictor importance locally and globally; and predictor interactions across subpopulations within distinctive life history contexts. Our results demonstrate a promising step toward interpretable machine learning applications to mid-to long-term prediction of travel modes for transportation planning.

Wi-Fi has become the wireless networking standard that allows short- to medium-range device to connect without wires. For the last 20 year, the Wi-Fi technology has so pervasive that most devices in use today are mobile and connect to the internet through Wi-Fi. Unlike wired network, a wireless network lacks a clear boundary, which leads to significant Wi-Fi network security concerns, especially because the current security measures are prone to several types of intrusion. To address this problem, machine learning and deep learning methods have been successfully developed to identify network attacks. However, collecting data to develop models is expensive and raises privacy concerns. The goal of this paper is to evaluate a federated learning approach that would alleviate such privacy concerns. This initial work on intrusion detection is performed in a simulated environment. Once proven feasible, this process would allow edge devices to collaboratively update global anomaly detection models, without sharing sensitive training data. On a set of tests with the AWID intrusion detection data set, we show that our federated approach is effective in terms of classification accuracy, computation cost, as well as communication cost.

We consider a science network that runs applications requiring data transfers to be completed within a given deadline. The underlying network is a software defined network (SDN) that supports fine grain real-time network telemetry. Deadline-aware data transfer requests are made to a centralized network controller that schedules the flows by setting pacing rates of the deadline flows and metering the background traffic at the ingress routers. The goal of the scheduling algorithm is to maximize the number of flows that meet the deadline while maximizing the network utilization. In this paper, we develop a Reinforcement Learning (RL) agent based network controller and compare its performance with well-known heuristics. For a network consisting of a single bottleneck link, we show that the RL-agent based network controller performs as well as Earliest Deadline First (EDF), which is known to be optimal. We also show that the RL-agent performs significantly better than an idealized TCP protocol in which the bottleneck link capacity is equally shared among the competing flows. We also study the sensitivity of the RL-agent controller for different parameter settings and reward functions.

For in transit processing, one of the fundamental challenges is the efficient movement of data from producers to consumers. Exploiting the flexibility offered by the SENSEI generic in situ framework, we have developed a number of different in transit data transport mechanisms. In this work, we focus on the transport mechanism that leverages the HDF5 parallel I/O library, and investigate the performance characteristics of this transport mechanism. For in transit use cases at scale on HPC platforms, one might expect that an in transit data transport mechanism that uses faster layers of the storage hierarchy, such as DRAM memory, would always outperform a transport that uses slower layers of the storage hierarchy, such as an NVRAM-based persistent storage presented as a distributed file system. However, our test results show that the performance of the transport using NVRAM is competitive with the transport that uses socket-based data movement across varying levels of producer and consumer concurrency.

Kui Wu,&ast; Hui-Hui Ke,&ast; Wei Gong, Hua Hu, Li Chen Department of Obstetrics and Gynecology, Shanghai Pudong Hospital, Fudan University Pudong Medical Center, Shanghai, 201399, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Li Chen; Hua Hu, Department of Obstetrics and Gynecology, Shanghai Pudong Hospital, Fudan University Pudong Medical Center, 2800 Gongwei Road, Shanghai, 201399, People’s Republic of China, Tel +86 21 5802 2995, Email chenli8369@163.com; Huhua20212022@163.comAim: To assess the impact of maternal pre-pregnancy body mass index (BMI) on gestational diabetes mellitus (GDM) based on different pre-pregnancy hemoglobin levels.Methods: This retrospective cohort study included 1289 pregnant women between June 2020 and January 2022. Clinical data were collected by reviewing their medical and antepartum screening records between 24 and 28 gestational weeks, including pre-pregnancy BMI and pre-pregnancy hemoglobin (Hb) levels. The diagnosis of GDM mainly depended on oral glucose tolerance test (OGTT) during 24– 28 weeks. Restricted cubic spline (RCS) was used to investigate the association between the pre-pregnancy Hb level and the risk of GDM. Univariate and multivariate logistic regression analyses were applied to evaluate the relative risk of GDM.Results: Of the 1289 included pregnant women, 187 (14.5%) women were diagnosed with GDM in this study. The pre-pregnancy Hb level was significantly associated with GDM risk, and the pre-pregnancy Hb level of 123 g/L was identified as the threshold to stratify and assess the association between the GDM risk and the pre-pregnancy BMI. For women with a pre-pregnancy Hb level ≥ 123 g/L, the pre-pregnancy BMI showed a significant association with GDM risk, and the estimated incidence rate of GDM was 7.7%, 14.8%, 36.3% and 44% for underweight, normal-weight, overweight and obese pregnant women, respectively. After adjusting for potential influencing factors of GDM, the respective relative risk was 1.0 (reference), 2.04 (95% CI 0.84, 4.99), 7.06 (2.66, 18.61), and 10.77 (2.85, 40.63) (P for trend < 0.001).Conclusion: In pregnant women with a pre-pregnancy Hb level ≥ 123 g/L, pre-pregnancy BMI showed a more significant association with GDM risk as compared with those with a pre-pregnancy Hb level < 123 g/L.Keywords: pregnancy, body mass index, hemoglobin, gestational diabetes mellitus, overweight, obese

Siyan Liu,1 Jing Li,2 Lin Gu,3 Kunzhe Wu,2 Hua Xing1 1Department of Breast Surgery, China-Japan Union Hospital of Jilin University, Changchun, People’s Republic of China; 2Department of Scientific Research Center, China-Japan Union Hospital of Jilin University, Changchun, People’s Republic of China; 3Breast Surgery, Jilin Province Tumor Hospital, Changchun, People’s Republic of ChinaCorrespondence: Hua Xing, Department of Breast Surgery, China-Japan Union Hospital of Jilin University, Xian Tai Street, No. 126, Changchun, Jilin, 130033, People’s Republic of China, Tel +86 139 4412 7842, Email xingh@jlu.edu.cnAbstract: Triple-negative breast cancer (TNBC) exhibits high recurrence and mortality rates because of the lack of effective treatment targets. Surgery and traditional chemotherapy are the primary treatment options. Immunotherapy shows high potential for treating various cancers but exhibits limited efficacy against TNBC as a monotherapy. Chemoimmunotherapy has broad prospects for applications for cancer treatment conferred through the synergistic immunomodulatory and anti-tumor effects of chemotherapy and immunotherapeutic strategies. However, improving the efficacy of synergistic therapy and reducing the side effects of multiple drugs remain to be the main challenges in chemoimmunotherapy against TNBC. Nanocarriers can target both cancer and immune cells, promote drug accumulation, and show minimal toxicity, making them ideal delivery systems for chemotherapeutic and immunotherapeutic agents. In this review, we introduce the immunomodulatory effects of chemotherapy and combined mechanisms of chemoimmunotherapy, followed by a summary of nanoparticle-mediated chemoimmunotherapeutic strategies used for treating TNBC. This up-to-date synthesis of relevant findings in the field merits contemplation, while considering avenues of investigation to enable advances in the field.Keywords: chemotherapy, drug-delivery system, immunotherapy, mechanism, targeted delivery

Fan Zhang,1,&ast; Jiayu Liang,1 You Lu,2,&ast; Yongquan Tang,3 Shengzhuo Liu,1 Kan Wu,1 Fuxun Zhang,1 Yiping Lu,1 Zhihong Liu,1 Xianding Wang1 1Department of Urology, Institute of Urology, West China Hospital, Sichuan University, Chengdu, 610041, People’s Republic of China; 2Department of Pediatrics, West China Second University Hospital, Chengdu, Sichuan University, Chengdu, 610041, People’s Republic of China; 3Department of Pediatric Urology, West China Hospital, Sichuan University, Chengdu, 610041, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xianding Wang; Jiayu Liang, Email xiandingwang66@126.com; drliangjiayu@163.comBackground: Cathepsin Z (CTSZ) is a cathepsin family member that plays a dual role in the adhesion and migration of immune and tumor cells.Methods: The expression pattern of CTSZ in clear cell renal cell carcinoma (ccRCC) was observed by immunohistochemistry and validated by using double-labeling immunofluorescence. Publicly available single-cell sequencing data was used to further define the cell type-specific CTSZ expression in ccRCC. Methylation modification, immune infiltration, and tumor-related signaling enrichment analyses involving CTSZ were performed using multi-omics data. Data from two independent cohorts of anti-programmed death-1 (PD-1) therapeutic clinical trials were used to investigate correlations between CTSZ levels and treatment responses.Results: CTSZ was upregulated in ccRCC tissues compared with adjacent normal tissues at the RNA but not in ccRCC cells. Immunohistochemistry indicated that CTSZ was expressed in tumors infiltrated with lymphocytes. Double immunofluorescence demonstrated that CTSZ was co-expressed with CD68 but not CD8. Single-cell transcriptome data showed macrophage-specific expression of CTSZ in ccRCC. High CTSZ expression was significantly correlated with the enrichment of interferon-γ, epithelial‐to‐mesenchymal transition, cell cycle, apoptosis pathways, and B cell, macrophage, neutrophil, and dendritic cell infiltrations, as well as the expression of immune checkpoints CTLA4, LAG3, HAVCR2, PDCD1LG2, PDCD1, TIGIT, and SIGLEC15. Hypomethylation modification of cg02744249, cg02744249, and cg22145559 were negatively correlated with CTSZ expression, suggesting an epigenetic mechanism for the regulation of CTSZ expression. Clinically, CTSZ levels were associated with the prognosis of patients with ccRCC (hazard ratio=1.5, P=0.007). Notably, patients with higher CTSZ expression had a worse prognosis with anti-PD-1 monotherapy (hazard ratio=1.51, P=0.039).Conclusion: Macrophage-specific CTSZ was associated with activation of epithelial‐to‐mesenchymal transition, cell cycle signatures, and a higher infiltration level of B cells, macrophages, neutrophils, and dendritic cells in the tumor microenvironment. High expression of CTSZ could be considered as a prognostic and treatment response biomarker for patients with ccRCC receiving anti-PD-1 immunotherapy.Keywords: ccRCC, CTSZ, macrophage, single-cell, prognosis

Limin Yan1 &ast;, Dedong Xu2 &ast;, Ying Xiao,1 Mingming Dai,1 Ting Wang,1 Xinhong Zhuang,3 Kunliang Wu4 1Department of Neurology, The Second Affiliated Hospital of Hainan Medical University, Haikou, People’s Republic of China; 2Department of Neurosurgery, The Second Affiliated Hospital of Hainan Medical University, Haikou, People’s Republic of China; 3Department of Nephrology, The Second Affiliated Hospital of Hainan Medical University, Haikou, People’s Republic of China; 4Department of Infectious Diseases, The Second Affiliated Hospital of Hainan Medical University, Haikou, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xinhong Zhuang, Department of Nephrology, The Second Affiliated Hospital of Hainan Medical University, No. 48, Baishuitang Road, Haikou, Hainan, 570311, People’s Republic of China, Tel +86-0890-66809015, Email zhuang_xinhong@163.com Kunliang Wu, Department of Infectious Diseases, The Second Affiliated Hospital of Hainan Medical University, No. 48, Baishuitang Road, Haikou, Hainan, 570311, People’s Republic of China, Tel +86-0890-66809015, Email kun4821583@163.comPurpose: NAD(P)H: Quinone Oxidoreductase 1 gene (NQO1) polymorphism is associated with the risk of cardiovascular disease. This study was designed to investigate the relationship between NQO1 gene polymorphism and ischemic stroke susceptibility in Chinese Han nationality.Patients and Methods: One hundred and forty-one patients diagnosed with ischemic stroke and 139 matched control groups were recruited in this study. The polymorphism distribution of rsl800566 locus and rs10517 locus of NQO1 gene was genotyped via TaqMan assay, and the concentration of Oxidized low-density lipoprotein (ox-LDL) in the blood of the subjects was detected by enzyme linked immunosorbent assay (ELISA). The relationship between the polymorphism distribution and the susceptibility to ischemic stroke was evaluated.Results: The frequency distribution of the three genotypes of NQO1 rs1800566 between the case group and the control group was statistically significant, and cases carrying CT and TT genotype were less likely to suffer from ischemic stroke. Compared with individuals carrying T allele, C allele carriers have higher risk of ischemic stroke. However, there was no significant difference in frequency distribution among the three genotypes of NQO1 rs10517 between controls and patients.Conclusion: The NQO1 rs1800566 C allele may be a novel marker associated with ischemic stroke susceptibility in Chinese Han population. Polymorphism of rsl800566 locus in NQO1 gene may be protective against ischemic stroke risk.Keywords: genetic polymorphism, ischemic stroke, NQO1, oxidized low-density lipoprotein

Jichan Shi,1,&ast; Gexin Gao,2,&ast; Zhijie Yu,3,&ast; Kaihuai Wu,4 Youquan Huang,5 Lian-Peng Wu,6 Zhengxing Wu,1 Xinchun Ye,1 Chaochao Qiu,1 Xiangao Jiang1 1Department of Infectious Disease, Wenzhou Central Hospital, The Dingli Clinical Institute of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, People’s Republic of China; 2Department of Nursing School, Wenzhou Medical University, Wenzhou, Zhejiang, 325000, People’s Republic of China; 3Department of Hematology, Wenzhou Key Laboratory of Hematology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, 325000, People’s Republic of China; 4Departments of Infectious Diseases, Taishun People’s Hospital, Wenzhou, 325000, People’s Republic of China; 5Departments of Infectious Diseases, Yongjia People’s Hospital, Wenzhou, 325000, People’s Republic of China; 6Department of Laboratory, Wenzhou Central Hospital, The Dingli Clinical Institute of Wenzhou Medical University, Wenzhou, Zhejiang, 325000, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xiangao Jiang, Department of Infectious Disease of Wenzhou Central Hospital, The Dingli Clinical Institute of Wenzhou Medical University, 252 Baili East Road, Lucheng District, Wenzhou, Zhejiang, 325000, People’s Republic of China, Tel +86 13587691822, Fax +86 577 88070007, Email jiang_xiangao@126.comBackground: Tuberculosis (TB) is still the single pathogen infectious disease with the largest number of deaths worldwide. The relationship that intestinal microbiota disorder and de novo fatty acid synthesis metabolism have with disease progression in multi-drug resistant TB (MDR-TB) has not yet been fully studied.Objective: To investigate the effects of long periods of MDR-TB, pre-extensively drug-resistant TB (pre-XDR-TB), or rifampicin-resistant TB (RR-TB) on gut microbiome dysbiosis and advanced disease.Methods: The sample was chosen between March 2019 and September 2019 in Wenzhou Central Hospital and comprised 11 patients with pre-XDR-TB, 23 patients with RR-TB, and 28 patients with MDR-TB. Healthy individuals were chosen as the control group (CK group). An overnight fast blood sample was drawn via venipuncture into tubes without anticoagulant. For analysis, 300 mg of faeces from patients from the same group was mixed and analysed using DNA extraction, NGS sequencing, and bioinformatics. A QIAamp Fecal DNA Mini Kit was used to isolate the DNA. The extracted DNA was stored at − 20°C.Results: Advanced TB was concurrent with an elevated level of the proportions of acetyl-CoA carboxylase (ACC1) to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and fatty acid synthase (FASN) to GAPDH in de novo fatty acids synthesis, and Eubacterium, Faecalibacterium, Roseburia, and Ruminococcus were increased significantly in RR-TB patients compared to healthy individuals, whereas their abundance in the pre-XDR-TB and MDR-TB groups showed little change in comparison with the control group. Proteobacteria levels were greatly increased in the RR-TB and MDR-TB patient groups but not in the patients with pre-XDR-TB or the healthy subjects. The pre-XDR-TB group exhibited alterations of the intestinal microbiome: coliform flora showed the highest abundance of Verrucomicrobiales, Enterobacteriales, Bifidobacteriales and Lactobacillales. De novo fatty acids synthesis was enhanced in patients and was associated with the gut microbiome dysbiosis induced by the antimicrobials, with Bacteroidetes, Bacteroidales, and Bacteroidaceae displaying the most important correlations on a phylum, order, and family level, respectively.Conclusion: The progression to advanced TB was observed to be a result of the interaction between multiple interrelated pathways, with gut–lung crosstalk potentially playing a role in patients with drug-resistant TB.Keywords: multi-drug resistant TB disease, gut commensal, microbial imbalance, de novo fatty acids synthesis, microbiome biosignature alterations

Chunkang Liu,1,&ast; Kunzhe Wu,2,&ast; Huan Gao,3 Jianyang Li,3 Xiaohua Xu3 1Department of Gastrointestinal Surgery, China-Japan Union Hospital of Jilin University, Changchun, People’s Republic of China; 2Department of Scientific Research Center, China-Japan Union Hospital of Jilin University, Changchun, People’s Republic of China; 3Department of Nephrology, China-Japan Union Hospital of Jilin University, Changchun, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xiaohua Xu, Email xu_xh@jlu.edu.cnAbstract: Diabetic nephropathy (DN), a severe microvascular complication of diabetes mellitus (DM), is the most common form of chronic kidney disease (CKD) and a leading cause of renal failure in end-stage renal disease. No currently available treatment can achieve complete cure. Traditional treatments have many limitations, such as painful subcutaneous insulin injections, nephrotoxicity and hepatotoxicity with oral medication, and poor patient compliance with continual medication intake. Given the known drawbacks, recent research has suggested that nanoparticle-based drug delivery platforms as therapeutics may provide a promising strategy for treating debilitating diseases such as DN in the future. This administration method provides multiple advantages, such as delivering the loaded drug to the precise target of action and enabling early prevention of CKD progression. This article discusses the development of the main currently used nanoplatforms, such as liposomes, polymeric NPs, and inorganic NPs, as well as the prospects and drawbacks of nanoplatform application in the treatment of CKD.Keywords: diabetic nephropathy, DN, nanoplatform, route of administration

Yachen Ji,&ast; Xiangjun Xu,&ast; Kangfei Wu, Yi Sun, Hao Wang, Yapeng Guo, Ke Yang, Junfeng Xu, Qian Yang, Xianjun Huang, Zhiming Zhou Department of Neurology, The First Affiliated Hospital of Wannan Medical College, Wuhu, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Xianjun Huang; Zhiming Zhou, Department of Neurology, The First Affiliated Hospital of Wannan Medical College, 2# East Zheshan Road, Wuhu, 241000, People’s Republic of China, Tel +86-25-80860124, Fax +86-25-84664563, Email doctorhuangxj@hotmail.Com; neuro_depar@hotmail.ComPurpose: The systemic immune inflammatory index (SII), as a new marker, is widely used to predict the disease prognosis. We investigated the predictive value of SII for malignant cerebral edema (MCE) and whether postoperative MCE mediates the relationship between SII and functional prognosis in patients undergoing endovascular thrombectomy (EVT).Patients and Methods: A total of 829 patients with anterior circulation large-vessel occlusive stroke (LVOS) were registered, and 675 (81.4%) met the inclusion criteria. We collected baseline data upon admission, including SII. Postoperative computed tomography was performed to assess the presence and grading of cerebral edema (CED), and MCE was defined as a CED score of 3. A good prognosis was defined as a modified Rankin Scale (mRS) score of 0– 2 at the 90-day follow-up.Results: A total of 132 patients developed MCE after EVT. The patients were divided into MCE and non-MCE groups, and univariate and multifactorial analyses were performed. Among these risk factors, an elevated SII was independently correlated with the occurrence of MCE. In addition, the receiver operating characteristic (ROC) curve was used to assess the predictive capability of SII levels for prognosis. The area under the ROC was 0.69, and the optimal critical value was 2.14. In addition, postoperative MCE may partially account for the poorer functional prognosis of patients with elevated SII (regression coefficient changed by 40.3%).Conclusion: The SII is an independent predictor of malignant brain edema after EVT. Postoperative MCE is partly the reason for the poorer prognosis in patients with elevated SII.Keywords: acute ischemic stroke, endovascular treatment, systemic immune inflammatory index, malignant cerebral edema

Sun F, Wu K, Yao Z, et al. Onco Targets Ther. 2020;13:6755–6765. We, the Editors and Publisher of OncoTargets and Therapy, have retracted the following article. Since publication, concerns have been raised about the integrity of the data in the article. This includes the duplication of images from Figures 5 and 6. Specifically, The image for Figure 5C, NC, has overlap with the image for Figure 6D, NC. When approached for an explanation, the authors did not respond to our queries, nor did they provide original data for their study. As verifying the validity of the published work is core to the integrity of the scholarly record, we are therefore retracting the article and the authors were notified of this. We have been informed in our decision-making by our editorial policies and COPE guidelines. The retracted article will remain online to maintain the scholarly record, but it will be digitally watermarked on each page as ‘Retracted’.