Search

Representation learning with small labeled data have emerged in many problems, since the success of deep neural networks often relies on the availability of a huge amount of labeled data that is expensive to collect. To address it, many efforts have been made on training sophisticated models with few labeled data in an unsupervised and semi-supervised fashion. In this paper, we will review the recent progresses on these two major categories of methods. A wide spectrum of models will be categorized in a big picture, where we will show how they interplay with each other to motivate explorations of new ideas. We will review the principles of learning the transformation equivariant, disentangled, self-supervised and semi-supervised representations, all of which underpin the foundation of recent progresses. Many implementations of unsupervised and semi-supervised generative models have been developed on the basis of these criteria, greatly expanding the territory of existing autoencoders, generative adversarial nets (GANs) and other deep networks by exploring the distribution of unlabeled data for more powerful representations. We will discuss emerging topics by revealing the intrinsic connections between unsupervised and semi-supervised learning, and propose in future directions to bridge the algorithmic and theoretical gap between transformation equivariance for unsupervised learning and supervised invariance for supervised learning, and unify unsupervised pretraining and supervised finetuning. We will also provide a broader outlook of future directions to unify transformation and instance equivariances for representation learning, connect unsupervised and semi-supervised augmentations, and explore the role of the self-supervised regularization for many learning problems., Comment: published in IEEE Transactions on Pattern Analysis and Machine Intelligence

Oncofetal protein SALL4 is critical for cancer cell survival. Targeting SALL4, however, is only applicable in a fraction of cancer patients who are positive for this gene. To overcome this limitation, we propose to induce a cancer vulnerability by engineering a partial dependency upon SALL4. Following exogenous expression of SALL4, SALL4-negative cancer cells became partially dependent on SALL4. Treatment of SALL4-negative cells with the FDA-approved hypomethylating agent 5-aza-2′-deoxycytidine (DAC) resulted in transient upregulation of SALL4. DAC pretreatment sensitized SALL4-negative cancer cells to entinostat, which negatively affected SALL4 expression through a microRNA, miRNA-205, both in culture and in vivo. Moreover, SALL4 was essential for the efficiency of sequential treatment of DAC and entinostat. Overall, this proof-of-concept study provides a framework whereby the targeting pathways such as SALL4-centered therapy can be expanded, sensitizing cancer cells to treatment by transient target induction and engineering a dependency. Significance: These findings provide a therapeutic approach for patients harboring no suitable target by induction of a SALL4-mediated vulnerability.

OBJECTIVE We aim to explore the capacity of perioperative pupillary variables to predict acute pain in the post-anesthesia care unit (PACU). METHODS Patients scheduled to undergo thoracic or abdominal surgery under general anesthesia between April 2021 and June 2021 were enrolled. We measured the pupil diameter, pupillary light reflex (PLR), and pupillary reflex dilatation 5 min before anesthesia induction (T1), 5 min after intubation (T2), at the end of anesthesia (T3), immediately before extubation (T4), and 5 min after extubation (T5). We assessed the early postoperative pain intensity in the PACU using Numeric Rating Scales (NRS) at recovery, 5 min after recovery, and 10 min after recovery. Logistic regression models were used to evaluate the association between perioperative pupillary variables and postoperative pain intensity. RESULTS Fifty-one patients were enrolled, 50 of whom were included in the final analysis. A total of 13 patients (26%) needed remedial analgesia in the PACU. Pupil parameters at T1, T2, T3, and T5 were not associated with NRS in the PACU. Multiple logistic regression models and receiver operating characteristic (ROC) curves indicated that only latency of PLR at T4 can predict postoperative acute pain. The ROC analysis showed that the cutoff value for latency of PLR at T4 was 0.29 s to discriminate between no pain and pain, and the area under the curve was 0.778 (95% CI 0.634-0.922, p = 0.002) with sensitivity 50.0% and specificity 91.7%. CONCLUSION The latency of PLR immediately before extubation may be a useful predictor for postoperative acute pain in the PACU.

Differentiable architecture search (DARTS) enables effective neural architecture search (NAS) using gradient descent, but suffers from high memory and computational costs. In this paper, we propose a novel approach, namely Partially-Connected DARTS (PC-DARTS), to achieve efficient and stable neural architecture search by reducing the channel and spatial redundancies of the super-network. In the channel level, partial channel connection is presented to randomly sample a small subset of channels for operation selection to accelerate the search process and suppress the over-fitting of the super-network. Side operation is introduced for bypassing (non-sampled) channels to guarantee the performance of searched architectures under extremely low sampling rates. In the spatial level, input features are down-sampled to eliminate spatial redundancy and enhance the efficiency of the mixed computation for operation selection. Furthermore, edge normalization is developed to maintain the consistency of edge selection based on channel sampling with the architectural parameters for edges. Theoretical analysis shows that partial channel connection and parameterized side operation are equivalent to regularizing the super-network on the weights and architectural parameters during bilevel optimization. Experimental results demonstrate that the proposed approach achieves higher search speed and training stability than DARTS. PC-DARTS obtains a top-1 error rate of 2.55 percent on CIFAR-10 with 0.07 GPU-days for architecture search, and a state-of-the-art top-1 error rate of 24.1 percent on ImageNet (under the mobile setting) within 2.8 GPU-days.

Rotator cuff injuries are a common clinical condition of the shoulder joint. Surgery that involves reattaching the torn tendon to its humeral head bony attachment has a somewhat lower success rate. The scar tissue formed during healing of the rotator cuff leads to poor tendon-related mechanical properties. To promote healing, a range of genetic interventions, as well as cell transplantation, and many other techniques have been explored. In recent years, the therapeutic promise of mesenchymal stem cells (MSCs) has been well documented in animal and clinical studies. Some data have suggested that MSCs can promote angiogenesis, reduce inflammation and cell proliferation and increase collagen deposition. These functions are likely paracrine effects of MSCs, particularly mediated through exosomes. Here, we review the use of MSCs-related exosomes in tissues and organs. We also discuss their potential utility for treating rotator cuff injuries, and explore the underlying mechanisms of their effects.

In this work, we aim to address the problem of human interaction recognition in videos by exploring the long-term inter-related dynamics among multiple persons. Recently, Long Short-Term Memory (LSTM) has become a popular choice to model individual dynamic for single-person action recognition due to its ability to capture the temporal motion information in a range. However, most existing LSTM-based methods focus only on capturing the dynamics of human interaction by simply combining all dynamics of individuals or modeling them as a whole. Such methods neglect the inter-related dynamics of how human interactions change over time. To this end, we propose a novel Hierarchical Long Short-Term Concurrent Memory (H-LSTCM) to model the long-term inter-related dynamics among a group of persons for recognizing human interactions. Specifically, we first feed each person's static features into a Single-Person LSTM to model the single-person dynamic. Subsequently, at one time step, the outputs of all Single-Person LSTM units are fed into a novel Concurrent LSTM (Co-LSTM) unit, which mainly consists of multiple sub-memory units, a new cell gate, and a new co-memory cell. In the Co-LSTM unit, each sub-memory unit stores individual motion information, while this Co-LSTM unit selectively integrates and stores inter-related motion information between multiple interacting persons from multiple sub-memory units via the cell gate and co-memory cell, respectively. Extensive experiments on several public datasets validate the effectiveness of the proposed H-LSTCM by comparing against baseline and state-of-the-art methods.

To evaluate the recovery of early urinary continence in patients with prostate cancer using a suprapubic catheter during Retzius-sparing robotic-assistant laparoscopic prostatectomy. From January 2018 to January 2019, 223 patients diagnosed with prostate cancer who underwent Retzius-sparing robotic-assistant laparoscopic prostatectomy in Diakonie Klinikum Stuttgart were involved in our study. From January 2018 to June 2018, patients (112 cases) only had an indwelling urinary catheter during Retzius-sparing robotic-assistant laparoscopic prostatectomy, while from July 2018 to January 2019, patients (111 cases) were offered an extra suprapubic catheter during operation. The recovery of early urinary continence of patients was mainly investigated one month later. The overall early urinary continence rate was 81.61%. Patients with suprapubic catheter had better urinary control results, compared to patients with only indwelling urinary catheter (87.39% vs 75.89%, p = 0.027). In addition, International Prostate Symptom Score and irritative subscore in patients with good urinary control were significantly lower than that in patients with urinary incontinence. Suprapubic catheter insertion (OR 0.395; 95% CI 0.190–0.821) and advanced pathological tumor stage (T3a–T4) (OR 2.061; 95% CI 1.008–4.217) were two independent influencing factors for early urinary continence recovery in patients who underwent Retzius-sparing robotic-assistant laparoscopic prostatectomy through multivariate logistic regression analysis. Suprapubic catheter insertion may be helpful for early urinary continence recovery in patients with Retzius-sparing Robotic-assistant laparoscopic prostatectomy. Advanced pathological tumor stage (T3a–T4) before Retzius-sparing robotic-assistant laparoscopic prostatectomy might be associated with poor urinary control.

Background: Hepatitis B virus infection is associated with liver disease, including cancers. In this study, we assessed the power of sex-determining region Y (SRY)-related high-mobility group (HMG)-box 4(SOX4) gene to predict the clinical course of hepatocellular carcinoma (HCC). Methods: To evaluate the differential expression of SOX4 and its diagnostic and prognostic potential in HCC, we analyzed the GSE14520 dataset. Stratified analysis and joint-effect analysis were done using SOX4 and clinical factor. We then designed a nomogram for predicting the clinical course of HCC. Differential SOX4 expression and its correlation with tumor stage as well as its diagnostic and prognostic value were analyzed on the oncomine and GEPIA websites. Gene set enrichment analysis was explored as well as candidate gene ontology and metabolic pathways modulated by in SOX4 HCC. Results: Our analysis revealed that the level of SOX4 was significantly upregulated in tumor issue (P

Despite advancements in treatment options, the overall cure and survival rates for non–small cell lung cancers (NSCLC) remain low. While small-molecule inhibitors of epigenetic regulators have recently emerged as promising cancer therapeutics, their application in patients with NSCLC is limited. To exploit epigenetic regulators as novel therapeutic targets in NSCLC, we performed pooled epigenome-wide CRISPR knockout screens in vitro and in vivo and identified the histone chaperone nucleophosmin 1 (Npm1) as a potential therapeutic target. Genetic ablation of Npm1 significantly attenuated tumor progression in vitro and in vivo. Furthermore, KRAS-mutant cancer cells were more addicted to NPM1 expression. Genetic ablation of Npm1 rewired the balance of metabolism in cancer cells from predominant aerobic glycolysis to oxidative phosphorylation and reduced the population of tumor-propagating cells. Overall, our results support NPM1 as a therapeutic vulnerability in NSCLC. Significance: Epigenome-wide CRISPR knockout screens identify NPM1 as a novel metabolic vulnerability and demonstrate that targeting NPM1 is a new therapeutic opportunity for patients with NSCLC.

Background This study was mainly to explore and study the potential application of lipoxygenases (ALOX) family genes in the diagnostic and prognostic values of colon adenocarcinoma (COAD). Methods Data sets related to the ALOX genes of COAD were obtained from The Cancer Genome Atlas and the University of California, Santa Cruz Xena browser. Then, the relevant biological information was downloaded from the public data platform. Finally, the bioinformatics technologies and clinical verification were employed to comprehensively analyze the potential values of ALOX genes. Results The Pearson correlation analysis indicated that there were correlations among ALOXE3, ALOX5, ALOX12, and ALOX12B. The diagnostic receiver operating characteristic (ROC) curves suggested that ALOXE3 and ALOX12 had significant diagnosis in COAD: ALOXE3; P

Representation learning methods allow a system to automatically learn robust and discriminative features from raw data for given goals, which play an important role in various visual content understanding applications, such as visual object segmentation, detection, tracking, recognition, and search. The performance of visual content understanding tasks is heavily dependent on the choice of data representation (or features) on which they are applied. Conventional feature representation methods usually employ transformations of data that make it easier to extract useful information, such as scale-invariant feature transform (SIFT), local binary patterns (LBP), and histogram of oriented gradients (HOG). In recent years, deep learning techniques have been widely applied to learn data-driven representations with supervised annotations and achieved great success in different visual content understanding tasks. Representative methods include the ResNet method for image classification, the DeepFace method for face recognition, and the feature pyramid networks (FPNs) method for object detection. Despite recent progresses on deep representation learning with a great amount of annotated data, how to effectively learn visual representation with limited data annotations still requires many efforts. This special section focuses on data-effective representation learning methods for visual content understanding.

OBJECTIVES To validate the operational and diagnostic performances of a new device for transient elastography (TE), FibroTouch, for liver fibrosis in patients with chronic hepatitis B (CHB). METHODS In this prospective multicenter study, adult patients with CHB and valid liver pathological results were recruited to validate the operational and diagnostic performance of a TE device by FibroTouch for staging liver fibrosis. RESULTS In total, 517 patients with histologically proven CHB were enrolled. All had achieved at least 10 successful liver stiffness measurements (LSM), resulting in a success rate of 99.1% and reliable evaluations of 95.2%. Altogether 412 patients were included to analyze the diagnostic performance of FibroTouch. The area under the receiver operating characteristic curve for the LSM was 0.846 (95% confidence interval [CI] 0.808-0.880) for fibrosis stage ≥ F1, 0.850 (95% CI 0.811-0.883) for ≥ F2, 0.908 (95% CI 0.876-0.934) for ≥ F3 and 0.874 (95% CI 0.836-0.903) for F4. The diagnostic accuracy of LSM was superior to that of gamma-glutamyl transpeptidase-to-platelet ratio (GPR), aminotransferase-to-platelet ratio index (APRI), or fibrosis index based on 4 factors (FIB-4) index in staging fibrosis F2-F4 (P = 0.007 to

Immune-based rather than cancer-based classification systems are becoming an important additional component for prognostic prediction. Herein, we investigate the status of tumor-immune interaction and prognostic value of immune activity scores in bladder urothelial carcinoma (BLCA), kidney renal clear cell carcinoma (KIRC), and prostate adenocarcinoma (PRAD) from The Cancer Genome Atlas dataset and tracking the tumor immunophenotype platform. Traditional clinicopathological parameters still are effective predictors for prognosis. Comparison of 23-set, seven-step immune activity scores, and its distribution between favorable and adverse outcome groups within each cancer type and among cancer types show that each cancer has a distinct tumor-immune pattern. Adjusted overall immune activity scores based on the binary logistic regression analysis show a great discrimination ability in the progression-free survival (p = .0056 in BLCA, p < .0001 in KIRC, and p < .0001 in PRAD). K-mean cluster method and principal component analysis were performed to explore the cancer subtype. The results reveal that certain immune activity scores pattern such as high nature killer (NK) cell, T cell, macrophage, T helper type 1 (Th1) cell, and dendritic cell recruiting scores in BLCA, low NK cell, CD8 T cell, macrophage, T cell, Th1 cell recruiting scores in KIRC have a favorable clinical outcome comparing to other subtypes. Also, It is remarkable that a relatively small subtype of patients has a poor clinical prognosis in PRAD and this subtype features high CD4 T cell, macrophage, T cell, and regulatory T cell recruiting scores.

Recent emergence of “microblogging” services has been driving cyber-physical social system (CPSS) as a hot topic in real-world applications. How to efficiently detect and recognise spam and fake accounts becomes an important task where it requires analysis of microblog user behavior and prediction of their activity. This article attempts to investigate this challenge by proposing a new strategy to effectively model microblogging user activity and dynamically predicting their activities for the CPSS applications. We first analysis and define a set of benchmarks for measuring microblogging user activeness in considering serval key dynamic attributes including change rate of microblogging numbers, user attentions, etc. Then, we build up a new dynamic microblogging user activity prediction model (DUAPM) based on three important characteristics: personal information, social relationship, and user interaction. Finally, an improved logical regression algorithm is proposed for training the model and predicting user activity. Under the evaluation of a sample dataset containing Sina Weibo 3621 users over 20 weeks, it shows that our model deliver average up to 3% higher prediction accuracy than other social media user activity prediction models using traditional logical regression and random forest algorithms. We also take out a CPSS case study of evaluating DUAPM models for analysis and prediction of Twitter users’ activity over 16 countries. The results show that our model effectively reflects the distribution and trends of Twitter users’ activity with different background and cultures.

Background This meta-analysis aims to systematically evaluate the evidence for mindfulness-based stress reduction (MBSR) in cancer related fatigue (CRF). Material and methods In October 2018, PubMed, Embase, Cochrane Library, Clinical Trials, China Biomedical Literature Database (CBM), China National Knowledge Infrastructure (CNKI) and China Science Periodical Database (CSPD) were searched for randomized controlled trials on MBSR in CRF patients. Literature screening and data extraction were conducted by two reviewers. Methodological quality evaluation was assessed by the Cochrane risk of bias tool. Revman 5.3.0 performs data analysis. The trial sequential analysis software estimated the required information size for each outcome indicator. Results There have been 5 studies included in this research for meta-analysis, 356 cases in the experimental group and 344 cases in the control group. The meta-analysis result indicates that: MBSR can reduce the cancer-related fatigue score of cancer patients, SMD = −0.51,95%CI [−0.81–0.20], P = 0.001, and the difference is statistically significant. The trial sequential analysis indicates that: The RIS required for the indicator to reach the level of significance test should be 1768. The sample size (700 cases) included in the study has not reached the RIS, but it has crossed the traditional threshold and the TSA threshold, indicating that the results tend to be stable. The grading results are shown as low-quality evidence. Conclusions This research has used evidence-based medicine to evaluate whether MBSR can alleviate CRF in cancer patients and provide evidence for the comprehensive intervention program for patients with cancer-related fatigue.

Background and Aim In China, clinical experience with direct‐acting antiviral treatments for hepatitis C virus (HCV) infection is still emerging. C‐CORAL is a phase 3, multinational, placebo‐controlled, double‐blind trial of elbasvir/grazoprevir (EBR/GZR) in participants with HCV infection from the Asia‐Pacific region and Russia. Here, we report the data from participants enrolled in China. Methods Treatment‐naive participants with chronic HCV genotype (GT) 1, GT4, or GT6 infection were randomly assigned to receive 50 mg EBR/100 mg GZR for 12 weeks (immediate‐treatment group, ITG) or placebo followed by deferred treatment with EBR/GZR (deferred‐treatment group, DTG). The primary efficacy end‐point was sustained virologic response at 12 weeks after completing treatment (SVR12), and the primary safety end‐point was a comparison of safety between participants receiving EBR/GZR and placebo (NCT02251990; Protocol PN‐5172‐067). Results A total of 152 participants in China were randomly assigned (ITG, n = 115; DTG, n = 37). SVR12 was achieved in 96.7% (146/151) participants overall and in 97.3% (142/146) of those with GT1b infection. Four participants relapsed (GT1b, n = 3; GT6a, n = 1). Drug‐related AEs were reported in 25 (21.7%) and 9 (24.3%) participants receiving EBR/GZR and placebo, respectively; no drug‐related serious adverse events (AEs) occurred. Two (1.7%) participants receiving EBR/GZR had late hepatic transaminase elevations. Patient‐reported outcomes indicate improved quality of life at follow‐up week 4 in participants receiving EBR/GZR compared to placebo. Conclusion EBR/GZR administered for 12 weeks represents a highly effective and safe treatment option for Chinese individuals with HCV GT1 infection., A sustained virologic response was achieved by 96.7% (146/151) of treatment‐naive participants with hepatitis c virus infection receiving elbasvir/grazoprevir (50 mg/100 mg) for 12 weeks. Patient‐reported outcomes indicate improved quality of life following clearance of HCV.

Intellectual property rights encourage people to facilitate the advance of industry and culture and drive the development of economic activity through creation. For this reason, the issue of intellectual property, under the environment, has become the basic knowledge in people’s life. With the protection of law, people are willing to contribute the intelligence and creation to the society for the innovation and advance of the country. With experimental design, total 98 students in two classes of a university in Guangxi are selected for the experimental research. Internet synchronous teaching is utilized for the 16-week (3 hours per week for total 48 hours) experimental teaching. The research results reveal positive relationship between 1.Internet synchronous teaching and learning motivation, 2.learning motivation and learning outcome, and 3.Internet synchronous teaching and learning outcome. According to the results to proposed suggestions, it is expected to have learners acquire the basic knowledge and idea of intellectual property rights as well as learn the information related to intellectual property rights or precede exchange and discussion through Internet synchronous teaching to enhance students’ impression and further achieve the digital learning of intellectual property rights.

Background To investigate the safety and efficacy of scleral concave pool trabeculectomy (SCPT) combined phacoemulsification for eyes with coexisting cataract and primary open-angle glaucoma (POAG). Methods This was a retrospective, controlled, interventional pilot case series. Thirty patients (30 eyes) were diagnosed with coexisting cataract and POAG between May 2015 and April 2018. Fourteen eyes underwent SCPT combined phacoemulsification were set as the study group, and 16 eyes received conventional phacotrabeculectomy were set as the control group. All patients were followed up for at least 6 months. The preoperative to postoperative changes in IOP, glaucoma medication requirements, BCVA, blebs functions, and adverse events were recorded. Results The groups were matched for baseline age, BCVA, IOP and types of IOP-lowering medications (all P > 0.05). At 6-month visit, there were no significant difference between control and study group in the improvement of BCVA (0.22 ± 0.24 versus 0.18 ± 0.26, P = 0.718), reduction of IOP (− 11.21 ± 8.61 mmHg versus − 9.19 ± 9.18 mmHg, P = 0.540) and the number of eyes that needed IOP-lowering medications (2 versus 3, P = 0.743). At the last visit, the rate of forming functioning blebs was significantly different between the study and control groups, (92.9% versus 68.7% respectively, P = 0.007). In the study group, 5 eyes developed hypotony, and 1 eye showed limited choroidal detachment, whereas in the control group 1 eye developed malignant glaucoma. All adverse events were successfully managed. Conclusion The SCPT combined phacoemulsification seems to be a safe and effective alternative to conventional phacotrabeculectomy for patients with POAG and visually significant cataract in the short-term.

Background Although circulating tumor cells (CTCs) are considered as a surrogate marker in monitoring disease progression and treatment response in late stage prostate cancer (PCa), its clinical impact in localized PCa remains unclear, indicating the limitation that is simply based on cell count. This perspective observational study aimed to detect the epithelial-to-mesenchymal transition (EMT) subtypes of CTCs in localized PCa and analyze their clinical relevance and application in predicting PCa stages before surgery compared with the Partin table. Patients and methods Between August 2017 and April 2019, 80 newly diagnosed localized PCa patients were enrolled in the study. Peripheral blood samples (5 mL) were collected prior to surgery. The CanPatrolTM CTC enrichment technique, a size-based isolation method, was used to detect the EMT CTCs. Clinical relevance of the CTCs was analyzed with Spearman's rank correlation test. Models to predict pathological were built with multivariate logistic regression. Receiver operating characteristic (ROC) curve and area under the curve (AUC) analysis were performed to evaluate the accuracy of the prediction model. Results CTCs were detected in 55% of all patients. The biophenotypic CTCs were most valuable and closely correlated with PSA, Gleason score, D'Amico risk classification, and pathological stage in localized PCa. The mesenchymal subtype was rare in this population but associated with seminal vesicle invasion, while the epithelial subtype had limited clinical significance. In addition, the biophenotypic CTCs combined with traditional clinical variables were analyzed by multivariate logistic regression to predict organ-confined disease before surgery, of which the AUC reached 0.818 and was superior to the Partin table 2017 in our cohort. Conclusion This study highlights the clinical impact of the biophenotypic CTCs in localized PCa, which was most closely related to clinical variables and could help to predict pathology outcomes before surgery.

We aimed to verify the expression status and diagnostic significance of isocitrate dehydrogenase 1 (IDH1) in non‐small‐cell lung cancer (NSCLC), especially during early stages. Serum IDH1 levels were measured by ELISA. A total of 1223 participants (660 patients with NSCLC, 276 healthy controls [HCs], 95 patients with benign pulmonary conditions [BPCs], 135 patients with other cancers [OCs], and 57 samples with interfering factors) were divided into a training cohort and a validation cohort according to 3 testing centers. The IDH1 concentrations in the NSCLC group were obviously higher than those in the control groups (P, This is the first large‐scale study aiming to verify the level and diagnostic value of serum isocitrate dehydrogenase 1 (IDH1) in non‐small‐cell lung cancer (NSCLC) with well‐designed controls in multicenter‐based cohorts. IDH1 showed high diagnostic efficiency in early stage NSCLC and high specificity for NSCLC comparing to other cancers.

In visual object tracking, by reasonably fusing multiple experts, ensemble framework typically achieves superior performance compared to the individual experts. However, the necessity of parallelly running all the experts in most existing ensemble frameworks heavily limits their efficiency. In this paper, we propose POST, a POlicy-based Switch Tracker for robust and efficient visual tracking. The proposed POST tracker consists of multiple weak but complementary experts (trackers) and adaptively assigns one suitable expert for tracking in each frame. By formulating this expert switch in consecutive frames as a decision-making problem, we learn an agent via reinforcement learning to directly decide which expert to handle the current frame without running others. In this way, the proposed POST tracker maintains the performance merit of multiple diverse models while favorably ensuring the tracking efficiency. Extensive ablation studies and experimental comparisons against state-of-the-art trackers on 5 prevalent benchmarks verify the effectiveness of the proposed method.

Due to importantly beneficial effects on physical and mental health and strong association with many rehabilitation programs, Physical Activity Recognition and Measure (PARM) has been widely recognised as a key paradigm for a variety of smart healthcare applications. Traditional methods for PARM relies on designing and utilising Data fusion or machine learning techniques in processing ambient and wearable sensing data for classifying types of physical activity and removing their uncertainties. Yet they mostly focus on controlled environments with the aim of increasing types of identifiable activity subjects, improved recognition accuracy and measure robustness. The emergence of the Internet of Things (IoT) enabling technology is transferring PARM studies to an open and dynamic uncontrolled ecosystem by connecting heterogeneous cost-effective wearable devices and mobile apps and various groups of users [35] . Little is currently known about whether traditional Data fusion techniques can tackle new challenges of IoT environments and how to effectively harness and improve these technologies. In an effort to understand potential use and opportunities of Data fusion techniques in IoT enabled PARM applications, this paper will give a systematic review, critically examining PARM studies from a perspective of a novel 3D dynamic IoT based physical activity collection and validation model. It summarized traditional state-of-the-art data fusion techniques from three plane domains in the 3D dynamic IoT model: devices, persons and timeline. The paper goes on to identify some new research trends and challenges of data fusion techniques in the IoT enabled PARM studies, and discusses some key enabling techniques for tackling them.

AIM Genetic factors are believed to be implicated in the pathogenesis of relapsing polychondritis (RP). However, the molecular genetic determinants remain to be elucidated. This study aimed to detect the susceptibility genes of RP with whole-exome sequencing (WES) in a Chinese family and deepen our understanding of the pathogenesis of RP thereafter. METHOD A 32-year-old Chinese female proband with RP and her family including her mother with RP were enrolled in the study. The genomic DNA of 6 human subjects was extracted from peripheral blood and then gene allele mutations were identified using WES. Candidate variants with low frequency (

Background: It has been demonstrated that miRNAs play critical roles in the tumorigenesis and progression of various tumors.Objective: The purpose of this research was to determine the serum miR-63...

Precast concrete sandwich panels (PCSPs) have been extensively used as a facade wall in buildings to improve thermal efficiency. The connectors that link the two concrete wythes are mainly used for in-plane shear transfer. In this study, two cross-shaped glass-fibre-reinforced polymer connectors with different section areas are proposed, with the aim of achieving optimal performance and easy construction in PCSPs. First, the pull-out and shear capacities of the connectors were evaluated. Then four PCSPs and two reinforced concrete solid panel counterparts were fabricated and tested under out-of-plane static loads in order to evaluate their flexural behaviour. The parameters investigated included the connector section area and spacing. Moreover, two-dimensional finite-element (FE) analysis was conducted to reproduce the test. The results indicate that: (a) the proposed connectors can provide considerable performance in terms of anchorage and shear capacity; (b) the failure mode of the PCSP specimens was governed by connector failure; (c) the degree of composite action in terms of ultimate strength can be obviously enhanced by increasing the connector section area and/or reducing its spacing; and (d) based on a comparison with the test results, the proposed FE model can predict the flexural behaviour of the tested PCSPs.

To investigate the dry eye symptoms after cataract surgery in MGD patients and their relationships METHODS: The study included 115 patients (115 eyes) with age-related cataract that underwent uncomplicated cataract surgery, and the patients were divided into two groups according to the MGD diagnostic criteria: group A (MGD group) and group B (control group). Schirmer I test (ST-I), tear breakup time (TBUT), and corneal fluorescein staining (CFS) were performed preoperatively and at 3 days, 7 days, 14 days, and 30 days postoperatively. We also measured eyelid meibomian gland morphology, meibomian gland expression, and meibum character scores before and after the cataract surgery.Postoperatively, in group A, TBUT decreased and CFS scores increased significantly. ST-I increased in the early postoperative course but decreased later. The eyelid margin morphology scores and meibomian gland expression scores of group A significantly increased after the cataract operation. Thus, patients with MGD may have a greater chance of developing the dry eye disease after cataract surgery. Cataract surgery may aggravate the signs of MGD, and the severity of MGD may positively correlate with TBUT, CFS, and corneal lesions after surgery.The characteristics of dry eye after cataract surgery in patients with MGD are different from common cataract patients, changes in the early postoperative phase to the ocular surface were caused by surgical factors, and the damages to epithelial function in the later postoperative phase were mainly associated with the inflammation of the meibomian gland and eyelid.

The state-of-the-art machine learning approaches are based on classical von Neumann computing architectures and have been widely used in many industrial and academic domains. With the recent development of quantum computing, researchers and tech-giants have attempted new quantum circuits for machine learning tasks. However, the existing quantum computing platforms are hard to simulate classical deep learning models or problems because of the intractability of deep quantum circuits. Thus, it is necessary to design feasible quantum algorithms for quantum machine learning for noisy intermediate scale quantum (NISQ) devices. This work explores variational quantum circuits for deep reinforcement learning. Specifically, we reshape classical deep reinforcement learning algorithms like experience replay and target network into a representation of variational quantum circuits. Moreover, we use a quantum information encoding scheme to reduce the number of model parameters compared to classical neural networks. To the best of our knowledge, this work is the first proof-of-principle demonstration of variational quantum circuits to approximate the deep $Q$-value function for decision-making and policy-selection reinforcement learning with experience replay and target network. Besides, our variational quantum circuits can be deployed in many near-term NISQ machines., Accepted for publication by IEEE Access

Objective: The goal of our current study is to assess the immunohistochemical of p53, p21, nm23, and VEGF expression in hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) prognosis after hepatectomy, as well as the prospective molecular mechanisms of prognostic indicator. Methods: There were 419 HBV-related HCC patients who were from southern China of Guangxi province and were used to evaluate the immunohistochemical expression for these biomarkers in prognosis. A genome-wide expression microarray dataset of HBV-related HCC were obtained from GSE14520. Results: In our study, the expression of p53, p21, and nm23 in cancer tissues of patients with hepatitis B-related hepatocellular carcinoma did not affected the clinical outcome of 2 years, 5 years or overall. Patients with high expression of VEGF had a worse overall survival after 2 years of surgery than patients with low expression (adjusted P=0.040, adjusted HR = 1.652, 95% CI = 1.024-2.665). Survival analysis of VEGF in GSE14520 cohort also demonstrated that VEGF mRNA expression also significantly associated with HBV-related HCC OS (adjusted P=0.035, adjusted HR =1.651, 95% CI =1.035-2.634). The prospective molecular mechanisms by co-expression analysis suggested that VEGF might be correlated to regulation of cell proliferation, cell growth and apoptotic process, Rap1 signaling pathway, HIF-1 signaling pathway, PPAR signaling pathway, cell cycle. Whereas the GSEA suggested that VEGF might involve in the regulation of HIF and HIF1A pathway, and TP53 regulation pathway. Conclusion: Our findings suggested that VEGF might be a prognostic indicator of HBV-related HCC, and we also identified the VEGF prospective molecular mechanisms through the whole genome co-expression and GSEA approaches.

Domestic waste classification was incorporated into legal provisions recently in China. However, relying on manpower to detect and classify domestic waste is highly inefficient. To that end, in this article, we propose a multimodel cascaded convolutional neural network (MCCNN) for domestic waste image detection and classification. MCCNN combined three subnetworks (DSSD, YOLOv4, and Faster-RCNN) to obtain the detections. Moreover, to suppress the false-positive predicts, we utilized a classification model cascaded with the detection part to judge whether the detection results are correct. To train and evaluate MCCNN, we designed a large-scale waste image dataset (LSWID), containing 30 000 domestic waste multilabeled images with 52 categories. To the best of our knowledge, the LSWID is the largest dataset on domestic waste images. Furthermore, a smart trash can is designed and applied to a Shanghai community, which helped to make waste recycling more efficient. Experimental results showed a state-of-the-art performance, with an average improvement of 10% in detection precision.

Introduction: Targeted medication, including mostly biologics and small-molecule chemical drugs, is an important therapy for ankylosing spondylitis (AS). There are still limited data on the preference of different targeted drugs in Chinese AS patients.Methods: A questionnaire-based cross-sectional study was performed on AS patients from six hospitals in three provinces in South China. Anti-rheumatic diseases’ medication history includes the recent and previous usage of biologics or Janus kinase inhibitors (JAKi) in the last complete course of treatment, disease severity, and reasons for targeted-treatment change or preference.Results: 354 of 366 participants responded to the online survey. The participants’ median age was 32 years, with a median of 7.3 years of disease duration; 79.7% were male. 63.6% of them were in the course of biologics or JAKi. Generic ETN is the most widely used and willing-to-use biologic though the proportion of its usage shrunk in the present compared with the past. The choice of original-branded ADA demonstrated an increase in usage. The preference of secukinumab and tofacitinib depicted a quick ascending trend.Conclusion: TNF-α inhibitors (TNFi) are still the most popular targeted medication for AS in China. Their price influences patients’ preferences mostly. The doctor’s recommendation is also part of the equation. Rheumatologists should pay more attention to patients’ education to formulate targeted therapeutic plans.

BACKGROUND Human leucocyte antigen (HLA) class I genes and haplotypes correlate with hepatitis B virus (HBV) infection. Occult HBV infection (OBI), a special type of chronic HBV infection, is defined as HBV surface antigen negative patients with or those without serologic markers by the means of HBV DNA detection in human plasma or in liver tissue by a diagnostic test. So far, the associations of HLA I haplotypes with OBI have not been reported previously in China. METHODS A case-control study between 107 OBI subjects and 280 healthy controls from blood donors in the Blood Center of the Shaanxi Province was conducted in the present association analysis. The HLA-A, -B and -C loci of case-control subjects were detected and genotyped by polymerase chain reaction-sequence based typing. The HLA-A, -B and -C haplotypic frequencies were calculated by the maximum likelihood method. RESULTS The HLA-A*33:03-C*07:01G (pc = 0.039, odds ratio [OR] = 8.996, 95% confidence interval [CI] = 1.825-44.338), B*44:03-C*07:01G (pc = 0.0069, OR = 12.000, 95% CI = 2.507-57.436) and A*33:03-B*44:03-C*07:01G (pc = 0.04, OR = 7.094, 95% CI = 1.387-36.288) haplotypes showed a a significant positive association with OBI. Independent effects demonstrated that HLA-B*44:03 and HLA-C*07:01G gave the main contribution to risk, whereas HLA-A*33:03 was associated only by linkage disequilibrium. CONCLUSIONS This present study is the first to demonstrate that HLA I haplotypes are associated with OBI in the Shaanxi Han population. The present results suggest that HLA-B*44:03-C*07:01G might be a potential risk factor for OBI. Comparisons of the frequencies of HLA I haplotypes at high resolution were made between OBI from northern Chinese Han and controls. Frequencies of HLA-A-B, A-C, B-C, A-B-C between the OBI group (gray column) and the control group (blank column) were compared. Haplotypes with p

As a new revolution of Ubiquitous Computing and Internet of Things, multimodal wearable intelligence technique is rapidly becoming a new research topic in both academic and industrial fields. Owning to the rapid spread of wearable and mobile devices, this technique is evolving healthcare from traditional hub-based systems to more personalised healthcare systems. This trend is well-aligned with recent “Healthcare 4.0” which is a continuous process of transforming the entire healthcare value chain to be preventive, precise, predictive and personalised, with significant benefits to elder care. But empowering the utility of multimodal wearable intelligence technique for elderly care like people with dementia is significantly challenging considering many issues, such as shortage of cost-effective wearable sensors, heterogeneity of wearable devices connected, high demand for interoperability, etc. Focusing on these challenges, this paper gives a systematic review of advanced multimodal wearable intelligence technologies for dementia care in Healthcare 4.0. One framework is proposed for reviewing the current research of wearable intelligence, and key enabling technologies, major applications, and successful case studies in dementia care, and finally points out future research trends and challenges in Healthcare 4.0.

Background and Aims Esophageal squamous cell cancer (ESCC) is one of the leading malignant cancers with a high incidence and mortality. Exploring novel serum biomarkers will help improve the management and monitoring of ESCC. Methods In the present study, we first used a ProcartaPlex Array to screen for serum proteins that were increased in 40 ESCC patients compared with matched normal controls; we found that eight proteins (IL‐2, IL‐5, IP‐10, IL‐8, eotaxin, TNF‐α, HGF, and MIP‐1b) had higher serum levels in ESCC patients than in normal controls. We further verified the clinical relevance of the candidate biomarkers with a larger sample of sera. Results In the 174 tested ESCC patients and 189 normal controls, the serum levels of eotaxin and IP‐10 were significantly higher in patients than in normal controls (p = 0.0038, 0.0031). In particular, these two proteins were also elevated in the sera of patients with early‐stage (0‐IIA) ESCC (p = 0.0041, 0.0412). When combining CEA and CYFRA21‐1 (in use clinically) with eotaxin or IP‐10, the effectiveness of detecting ESCC was superior to that of CEA and/or CYFRA21‐1 alone. Moreover, the serum level of eotaxin dropped significantly after surgical resection of primary tumors compared with that in preoperative ESCC samples (p, Serum samples from 40 ESCC patients and 40 normal controls were used to screen protein candidates by using ProcartaPlex Array in the biomarker discovery stage. The diagnostic values of the candidate IL‐2, IL‐5, IP‐10, IL‐8, eotaxin, TNF‐α, HGF, MIP‐1b using ELISA were further evaluated in 363 serum samples in the biomarker assessment stage and were finally evaluation value of protein panels for the detection and therapy monitoring of ESCC.

We present a novel attention mechanism: Causal Attention (CATT), to remove the ever-elusive confounding effect in existing attention-based vision-language models. This effect causes harmful bias that misleads the attention module to focus on the spurious correlations in training data, damaging the model generalization. As the confounder is unobserved in general, we use the front-door adjustment to realize the causal intervention, which does not require any knowledge on the confounder. Specifically, CATT is implemented as a combination of 1) In-Sample Attention (IS-ATT) and 2) Cross-Sample Attention (CS-ATT), where the latter forcibly brings other samples into every IS-ATT, mimicking the causal intervention. CATT abides by the Q-K-V convention and hence can replace any attention module such as top-down attention and self-attention in Transformers. CATT improves various popular attention-based vision-language models by considerable margins. In particular, we show that CATT has great potential in large-scale pre-training, e.g., it can promote the lighter LXMERT~\cite{tan2019lxmert}, which uses fewer data and less computational power, comparable to the heavier UNITER~\cite{chen2020uniter}. Code is published in \url{https://github.com/yangxuntu/catt}.

As the rapid growth of social media technologies continues, Cyber-Physical-Social System (CPSS) has been a hot topic in many industrial applications. The use of “microblogging” services, such as Twitter, has rapidly become an influential way to share information. While recent studies have revealed that understanding and modelling microblog user behaviour with massive users’ data in social media are keen to success of many practical applications in CPSS, a key challenge in literatures is that diversity of geography and cultures in social media technologies strongly affect user behaviour and activity. The motivation of this article is to understand differences and similarities between microblogging users from different countries using social media technologies, and to attempt to design a Country-Level Micro-Blog User (CLMB) behaviour and activity model for supporting CPSS applications. We proposed a CLMB model for analysing microblogging user behaviour and their activity across different countries in the CPSS applications. The model has considered three important characteristics of user behaviour in microblogging data, including content of microblogging messages, user emotion index, and user relationship network. We evaluated CLBM model under the collected microblog dataset from 16 countries with the largest number of representative and active users in the world. Experimental results show that (1) for some countries with small population and strong cohesiveness, users pay more attention to social functionalities of microblogging service; (2) for some countries containing mostly large loose social groups, users use microblogging services as a news dissemination platform; (3) users in countries whose social network structure exhibits reciprocity rather than hierarchy will use more linguistic elements to express happiness in microblogging services.

Endoscopic transaxillary augmentation mammaplasty breast augmentation offers several advantages over other augmentation methods. Nonetheless, this procedure is fraught with some problems, including greater surgical trauma due to the longer separation area. We hypothesized that cold separation of the interpectoral space could reduce surgical injury in comparison to the electrosurgical method. This study aimed to compare the outcomes of endoscopic-assisted transaxillary augmentation mammaplasty using cold separation versus electrosurgical separation of the interpectoral space.In this prospective clinical trial, cold and electrosurgical separation of the interpectoral space were achieved using a separation shovel and monopolar electrotome, respectively. A total of 20 patients who visited our department in Beijing, China, for primary breast augmentation surgeries from October 1, 2017, and May 31, 2018, were included. The primary outcome was total postoperative drainage volume. The secondary outcomes were operative time, daily drainage volume, daily pain as assessed using the visual analogue scale (VAS), and reoperation rate. Quantitative data were compared using independent-samplesThe total drainage volume was significantly lower in the cold separation group than in the electrosurgical separation group (170.45 ± 75.40 mL vs 281.05 ± 148.43 mL;Cold separation is more conducive to reducing drainage, relieving postoperative pain, and causing less damage than the electrosurgical method in endoscopic-assisted transaxillary dual-plane augmentation mammaplasty.L’augmentation mammaire transaxillaire par voie endoscopique comporte plusieurs avantages par rapport aux autres méthodes d’augmentation. Cette intervention se heurte toutefois à certains problèmes, y compris des traumatismes chirurgicaux plus importants à cause de la zone de séparation plus longue. Les auteurs ont postulé que la séparation de l’espace interpectoral par le froid réduirait davantage la lésion que la méthode électrochirurgicale. La présente étude visait à comparer les résultats cliniques de l’augmentation mammaire transaxillaire assistée par endoscopie au moyen de la séparation par le froid plutôt que par la séparation électrochirurgicale de l’espace interpectoral.Dans la présente étude clinique prospective, la séparation par le froid et la séparation électrochirurgicale de l’espace interpectoral ont été assurées par une pince de séparation et une électrode monopolaire, respectivement. Au total, 20 participants qui ont consulté le département des auteurs à Beijing, en Chine, afin de subir une augmentation mammaire primaire entre le 1Le volume de drainage total était considérablement plus faible dans le groupe de séparation par le froid que dans celui de séparation électrochirurgicale (170,45 ± 75,40 mL par rapport à 281,05 ± 148,43 mL; P = 0,005). Le score d’ÉAV le premier jour postopératoire était considérablement plus faible dans le groupe de séparation par le froid que dans celui de séparation électrochirurgicale (6,45 ± 1,93 par rapport à 7,55 ± 1,43; P = 0,048). Deux réopérations (20 %) causées par la douleur postopératoire ou la rigidité de l’implant ont été exécutées dans le groupe de séparation électrochirurgicale.La séparation par le froid favorise la diminution du drainage, le soulagement de la douleur postopératoire et la réduction des dommages davantage que la méthode életrochirurgicale en cas d’augmentation mammaire transaxillaire biplan assistée par endoscopie.

A highly efficient and stable electrocatalyst with the novel heterostructure of Co-embedded and N-doped carbon nanotubes supported Mo2C nanoparticles (Mo2C/NCNTs@Co) is creatively constructed by adopting the one-step metal catalyzed carbonization-nitridation strategy. Systematic characterizations and density functional theory (DFT) calculations reveal the advanced structural and electronic properties of Mo2C/NCNTs@Co heterostructure, in which the Co-embedded and N-doped CNTs with tunable diameters present electron-donating effect and the work function is correspondingly regulated from 4.91 to 4.52 eV, and the size-controlled Mo2C nanoparticles exhibit Pt-like 4d electronic structure and the well matched work function (4.85 eV) with I-/I3- redox couples (4.90 eV). As a result, the conductive NCNTs@Co substrate with fine-tuned energy level alignment accelerates the electron transportation and the electron migration from NCNTs@Co to Mo2C, and the active Mo2C shows high affinity for I3- adsorption and high charge transfer ability for I3- reduction, which reach a decent synergetic catalytic effect in Mo2C/NCNTs@Co heterostructure. The DSSC with Mo2C/NCNTs@Co CE achieves a high photoelectric conversion efficiency of 8.82% and exceptional electrochemical stability with a residual efficiency of 7.95% after continuous illumination of 200 h, better than Pt-based cell. Moreover, the synergistic catalytic mechanism toward I3- reduction is comprehensively studied on the basis of structure-activity correlation and DFT calculations. The advanced heterostructure engineering and electronic modulation provide a new design principle to develop the efficient, stable, and economic hybrid catalysts in relevant electrocatalytic fields.

Background Calcifying fibrous tumor (CFT) is a rare benign mesenchymal tumor that often occurs in deep soft tissue of children and young adults. CFT rarely occurs in the mediastinum. Case summary In this paper, we describe a 31-year-old male patient with CFT in the mediastinum. The patient did not have any symptoms, and the posterior mediastinal lesion was unintentionally found during routine re-examination of thyroid cancer. The tumor had no adhesion to the surrounding tissue and was successfully and completely removed. Pathology showed a large amount of collagen-rich fibrous connective tissue. There was scattered dystrophic calcification and gravel in the fibrous tissue and a small amount of lymphocyte and plasma cell infiltration and lymphoid follicle formation in the interstitial fluid. In addition, findings showed 20 IgG4+ plasma cells per high-powered field of the diseased tissue, an IgG4+/IgG ratio of about 20%, and normal serum IgG4 levels. The final diagnosis was CFT of the mediastinum (CFTM). No evidence of tumor recurrence was observed by computed tomography at 3 mo after surgery. Conclusion IgG4+ plasma cell enlargement may occur in CFTM, but clinical manifestations and serological tests suggest that it is not IgG4-related disease. We speculate that it may be an independent tumor subtype.

Recent advancements in recurrent neural network (RNN) research have demonstrated the superiority of utilizing multiscale structures in learning temporal representations of time series. Currently, most of multiscale RNNs use fixed scales, which do not comply with the nature of dynamical temporal patterns among sequences. In this paper, we propose Adaptively Scaled Recurrent Neural Networks (ASRNN), a simple but efficient way to handle this problem. Instead of using predefined scales, ASRNNs are able to learn and adjust scales based on different temporal contexts, making them more flexible in modeling multiscale patterns. Compared with other multiscale RNNs, ASRNNs are bestowed upon dynamical scaling capabilities with much simpler structures, and are easy to be integrated with various RNN cells. The experiments on multiple sequence modeling tasks indicate ASRNNs can efficiently adapt scales based on different sequence contexts and yield better performances than baselines without dynamical scaling abilities.

BackgroundThe concept of mixed adenoneuroendocrine carcinoma (MANEC) was introduced in the 2010 World Health Organization classification of digestive neuroendocrine neoplasms. Bile duct invasion by MANEC is exceptionally rare. We herein report a case of MANEC with invasion of multiple bile ducts. Case presentation: A 60-year-old man presented with a 7-day history of upper abdominal pain, and a mass in the cystic duct was suspected based on computed tomography findings. The patient underwent resection of the extrahepatic bile ducts with concomitant radical lymphadenectomy and Roux-en-Y cholangiojejunostomy. Large cell neuroendocrine carcinoma was detected in a component of the resected tumor. According to the pathological and immunohistochemical features of the tumor, the final histopathological diagnosis was a biliary MANEC, tumor stage T2N0M1 (Stage IIIC). The patient recovered uneventfully and was discharged from the hospital 10 days after surgery.ConclusionsWe have described a rare case of extrahepatic MANEC invading multiple bile ducts, with particular emphasis on the physician’s awareness of MANEC and its optimal treatment. MANEC arising from extrahepatic bile ducts is rare, and surgical resection is the most effective treatment method.

In this article, we propose a novel deep Siamese architecture based on a convolutional neural network (CNN) and multi-level similarity perception for the person re-identification (re-ID) problem. According to the distinct characteristics of diverse feature maps, we effectively apply different similarity constraints to both low-level and high-level feature maps during training stage. Due to the introduction of appropriate similarity comparison mechanisms at different levels, the proposed approach can adaptively learn discriminative local and global feature representations, respectively, while the former is more sensitive in localizing part-level prominent patterns relevant to re-identifying people across cameras. Meanwhile, a novel strong activation pooling strategy is utilized on the last convolutional layer for abstract local-feature aggregation to pursue more representative feature representations. Based on this, we propose final feature embedding by simultaneously encoding original global features and discriminative local features. In addition, our framework has two other benefits: First, classification constraints can be easily incorporated into the framework, forming a unified multi-task network with similarity constraints. Second, as similarity-comparable information has been encoded in the network’s learning parameters via back-propagation, pairwise input is not necessary at test time. That means we can extract features of each gallery image and build an index in an off-line manner, which is essential for large-scale real-world applications. Experimental results on multiple challenging benchmarks demonstrate that our method achieves splendid performance compared with the current state-of-the-art approaches.

Many researches have brought progress in learning a good generative model by combing the advantages of GAN and VAE, where latent space learning is always important for generating high-quality images. But these existing works mainly seek to impose the latent space a given distribution in advance or to obey a Gaussian distribution with KL divergence penalty, which leads to the difficulties of deciding a suitable prior distribution corresponding to different datasets. Thus in this paper we develop a two-stage combining method of AE and GAN under unsupervised and supervised conditions respectively, each stage performed to improve the effect of modeling latent distribution. In the first stage, an adversarial procedure achieves to match the latent distribution with real data distribution determined by arbitrary dataset without having access to a pre-set prior distribution. In the second stage, besides one adversarial procedure trained for outputting images, the other adversarial procedure is designed to attain the goal of optimizing the latent distribution of the first stage via back-propagation. Therefore, loop optimization of the network parameters during training will at last allow the framework to map an input noise to a high quality image. Extensive experiments are conducted to verify the performance of representing latent space and generating images on different datasets including MNIST, fashion-MNIST, CIFAR-10, and CelebA. The code and tutorials already release at https://github.com/TwistedW/CAE-CGAN .

Hashing has attracted increasing research attention in recent years due to its high efficiency of computation and storage in image retrieval. Recent works have demonstrated the superiority of simultaneous feature representations and hash functions learning with deep neural networks. However, most existing deep hashing methods directly learn the hash functions by encoding the global semantic information, while ignoring the local spatial information of images. The loss of local spatial structure makes the performance bottleneck of hash functions, therefore limiting its application for accurate similarity retrieval. In this paper, we propose a novel deep ordinal hashing (DOH) method, which learns ordinal representations to generate ranking-based hash codes by leveraging the ranking structure of feature space from both local and global views. In particular, to effectively build the ranking structure, we propose to learn the rank correlation space by exploiting the local spatial information from fully convolutional network and the global semantic information from the convolutional neural network simultaneously. More specifically, an effective spatial attention model is designed to capture the local spatial information by selectively learning well-specified locations closely related to target objects. In such hashing framework, the local spatial and global semantic nature of images is captured in an end-to-end ranking-to-hashing manner. Experimental results conducted on three widely used datasets demonstrate that the proposed DOH method significantly outperforms the state-of-the-art hashing methods.

Real-time detection of circulating tumor cell (CTC) markers that are constantly changing and renewing during disease progression is of great significance for the timely regimen switch or individualized target therapy. The abnormally expressed special AT-rich sequence binding protein 1 (SATB1), a nuclear matrix attachment region binding protein, in various tumors, promotes the growth and metastasis of tumor cells by regulating gene expression. In this paper, a CTC detection system for prostate cancer (PCa) was developed on the basis of epithelial cell adhesion molecule (EpCAM)-targeted immunomagnetic separation and CK-FITC and SATB-1-APC immunofluorescence assay, and the recovery rate of tumor cells in PBS and simulated whole blood by this system was detected. Subsequently, we isolated, identified, and counted SATB-1 ositive CTCs in the peripheral blood and urine samples of 60 tumor-bearing nude mice, 5 healthy volunteers and 13 PCa patients. Combined with the clinicopathological factors, the clinical value of the system was analyzed, and the possibility of SATB-1-positive CTCs in the diagnosis of PCa was evaluated. The results showed that the CTC sorting and identification system for prostate cancer constructed in this study had a recovery rate of more than 85% for CTC in PBS, urine and blood simulation samples. The expression level of SATB-1 was different in different PCa cell lines, which was relatively high in the highly invasive PCa DU-145 cell line. The expression of SATB-1 in CTCs in the blood samples of PCa patients with different clinical characteristics and in the urine samples of a few PCa patients with bone metastases were different, and the detection sensitivity of peripheral blood was higher than that of urine. This study has important clinical reference value for the early diagnosis of PCa and the evaluation of bone metastasis based on the CTC counting and the SATB-1 expression in CTCs.

Patient outcomes for steroid-dependent or -refractory chronic graft-versus-host diesease (cGVHD) are poor, and only ibrutinib has been US Food and Drug Administration (FDA) approved for this indication. cGVHD is often driven by the germinal center (GC) reaction, in which T follicular helper cells interact with GC B cells to produce antibodies that are associated with disease pathogenesis. The transcriptional corepressor B-cell lymphoma 6 (BCL6) is a member of the Broad-complex, Tramtrack, and Bric-abrac/poxvirus and zinc finger (BTB/POZ) transcription factor family and master regulator of the immune cells in the GC reaction. We demonstrate that BCL6 expression in both donor T cells and B cells is necessary for cGVHD development, pointing to BCL6 as a therapeutic cGVHD target. A small-molecule BCL6 inhibitor reversed active cGVHD in a mouse model of multiorgan system injury with bronchiolitis obliterans associated with a robust GC reaction, but not in cGVHD mice with scleroderma as the prominent manifestation. For cGVHD patients with antibody-driven cGVHD, targeting of BCL6 represents a new approach with specificity for a master GC regulator that would extend the currently available second-line agents.

AIM: To evaluate the real-life clinical outcomes of intravitreal injection of conbercept combined rescue therapy for polypoidal choroidal vasculopathy (PCV). METHODS: This was an open label, single center, and interventional study. All enrolled patients were treated initially with three consecutive monthly intravitreal conbercept injections (0.5 mg). Additional conbercept injections were administered upon substantial polyp regression with improved visual acuity (VA). Eyes with partial or no polyp regression and poor VA were rescue treated with photodynamic therapy (PDT) for subfoveal polyps or thermal laser photocoagulation for extrafoveal polyps. Best-corrected visual acuity (BCVA), central foveal thickness (CFT) and polyp regression were observed as primary outcomes. Side effects were also collected during the follow-up period. RESULTS: A total of 56 eyes (56 patients) with PCV were included. BCVA increased significantly from the baseline of 43.52±24.21 letters to 55.88±21.94 letters (P

PGGFRα+ preadipocytes are the major subpopulations that can regenerate into adipocytes. Two different types of macrophages exist in the fat tissue: the classically activated macrophage (M1) and the alternatively activated macrophage (M2). In this study, we investigated whether M1/M2 macrophages play distinct roles in adipogenic differentiation of PDGFRα+ preadipocytes. Mouse preadipocytes and macrophages were isolated from C57BL/6 male mice of 6–8 weeks. The culture supernate of M1 and M2 macrophages was collected and co-cultured with the PDGFRα+ preadipocytes. After 3 days, Oil Red O staining was used to evaluate to adipogenic differentiation of PDGFRα+ preadipocytes and the expression of adipogenic-related transcription factors (C/EBP-α, PPARγ) were also tested. The results showed that when cultured in vitro, M1 macrophages could significantly suppress the adipogenesis of PDGFRα+ preadipocytes as well as the C/EBP-α and PPARγ expression, but M2 macrophages did not have significant influence on the adipogenesis of PDGFRα+ preadipocytes nor on C/EBP-α and PPARγ expression compared with the control group. M1 macrophages significantly suppress PDGFRα+ preadipocyte adipogenesis which provides a possible way to improve adipogenesis and fat retention after fat-free grafting by mitigating acute inflammation and manipulating M1 macrophage levels. This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

Large quantities of bacteria, including Firmicutes, Actinobacteria, and Bacteroidetes, colonize the surface of the respiratory mucosa of healthy people. They interact and coexist with the local mucosal immune system of the human airway, maintaining the immune stability and balance of the respiratory system. While suffering from chronic respiratory diseases, the microbial population in the airway changes and the proportion of Proteobacteria is increased in patients with asthma. The abundance of the microbial population in patients with chronic obstructive pulmonary disease (COPD) is decreased, and conversely, the proportion of Firmicutes and Proteobacteria increased. The diversity of airway microorganisms in cystic fibrosis (CF) patients is decreased, while pathogenic bacteria and conditional pathogenic bacteria are proliferated in large numbers. The proportion of Firmicutes and Proteobacteria is increased in patients with upper airway cough syndrome (UACS), which replaces the dominance of Streptococcus and Neisseria in the pharynx of a normal population. Therefore, a clear understanding of the immune process of the airway flora and the immune dysfunction of the flora on the pathogenesis of chronic respiratory diseases can provide new ideas for the prevention and treatment of human respiratory diseases.