Abstract: The social cost of carbon (SCC) is a central concept for understanding and implementing climate change policies. This term represents the economic cost caused by an additional ton of carbon dioxide emissions or its equivalent. The present study presents updated estimates based on a revised DICE model (Dynamic Integrated model of Climate and the Economy). The study estimates that the SCC is $31 per ton of CO2 in 2010 US$ for the current period (2015). For the central case, the real SCC grows at 3% per year over the period to 2050. The paper also compares the estimates with those from other sources.

Abstract: The aim of this study was to provide an overview of the recommendations regarding the diagnosis and treatment contained in current clinical practice guidelines for patients with non-specific low back pain in primary care. We also aimed to examine how recommendations have changed since our last overview in 2010. The searches for clinical practice guidelines were performed for the period from 2008 to 2017 in electronic databases. Guidelines including information regarding either the diagnosis or treatment of non-specific low back pain, and targeted at a multidisciplinary audience in the primary care setting, were considered eligible. We extracted data regarding recommendations for diagnosis and treatment, and methods for development of guidelines. We identified 15 clinical practice guidelines for the management of low back pain in primary care. For diagnosis of patients with non-specific low back pain, the clinical practice guidelines recommend history taking and physical examination to identify red flags, neurological testing to identify radicular syndrome, use of imaging if serious pathology is suspected (but discourage routine use), and assessment of psychosocial factors. For treatment of patients with acute low back pain, the guidelines recommend reassurance on the favourable prognosis and advice on returning to normal activities, avoiding bed rest, the use of nonsteroidal anti-inflammatory drugs (NSAIDs) and weak opioids for short periods. For treatment of patients with chronic low back pain, the guidelines recommend the use of NSAIDs and antidepressants, exercise therapy, and psychosocial interventions. In addition, referral to a specialist is recommended in case of suspicion of specific pathologies or radiculopathy or if there is no improvement after 4 weeks. While there were a few discrepancies across the current clinical practice guidelines, a substantial proportion of recommendations was consistently endorsed. In the current review, we identified some differences compared to the previous overview regarding the recommendations for assessment of psychosocial factors, the use of some medications (e.g., paracetamol) as well as an increasing amount of information regarding the types of exercise, mode of delivery, acupuncture, herbal medicines, and invasive treatments. These slides can be retrieved under Electronic Supplementary Material.

Abstract: We present parmbsc1, a force field for DNA atomistic simulation, which has been parameterized from high-level quantum mechanical data and tested for nearly 100 systems (representing a total simulation time of ~140 μs) covering most of DNA structural space. Parmbsc1 provides high-quality results in diverse systems. Parameters and trajectories are available at http://mmb.irbbarcelona.org/ParmBSC1/.

Abstract: Gastric cancer is the second most common cause of cancer-related deaths in the world, the epidemiology of which has changed within last decades. A trend of steady decline in gastric cancer incidence rates is the effect of the increased standards of hygiene, conscious nutrition, and Helicobacter pylori eradication, which together constitute primary prevention. Avoidance of gastric cancer remains a priority. However, patients with higher risk should be screened for early detection and chemoprevention. Surgical resection enhanced by standardized lymphadenectomy remains the gold standard in gastric cancer therapy. This review briefly summarizes the most important aspects of gastric cancers, which include epidemiology, risk factors, classification, diagnosis, prevention, and treatment. The paper is mostly addressed to physicians who are interested in updating the state of art concerning gastric carcinoma from easily accessible and credible source.

Abstract: A central goal in systems neuroscience is the parcellation of the cerebral cortex into discrete neurobiological “atoms”. Resting-state functional magnetic resonance imaging (rs-fMRI) offers the possibility of in-vivo human cortical parcellation. Almost all previous parcellations relied on one of two approaches. The local gradient approach detects abrupt transitions in functional connectivity patterns. These transitions potentially reflect cortical areal boundaries defined by histology or visuotopic fMRI. By contrast, the global similarity approach clusters similar functional connectivity patterns regardless of spatial proximity, resulting in parcels with homogeneous (similar) rs-fMRI signals. Here we propose a gradient-weighted Markov Random Field (gwMRF) model integrating local gradient and global similarity approaches. Using task-fMRI and rs-fMRI across diverse acquisition protocols, we found gwMRF parcellations to be more homogeneous than four previously published parcellations. Furthermore, gwMRF parcellations agreed with the boundaries of certain cortical areas defined using histology and visuotopic fMRI. Some parcels captured sub-areal (somatotopic and visuotopic) features that likely reflect distinct computational units within known cortical areas. These results suggest that gwMRF parcellations reveal neurobiologically meaningful features of brain organization and are potentially useful for future applications requiring dimensionality reduction of voxel-wise fMRI data. Multi-resolution parcellations generated from 1489 participants are available at FREESURFER_WIKI LINK_TO_BE_ADDED.

Abstract: Nature-based solutions promoting green and blue urban areas have significant potential to decrease the vulnerability and enhance the resilience of cities in light of climatic change. They can thereby help to mitigate climate change-induced impacts and serve as proactive adaptation options for municipalities. We explore the various contexts in which nature-based solutions are relevant for climate mitigation and adaptation in urban areas, identify indicators for assessing the effectiveness of nature-based solutions and related knowledge gaps. In addition, we explore existing barriers and potential opportunities for increasing the scale and effectiveness of nature-based solution implementation. The results were derived from an inter- and transdisciplinary workshop with experts from research, municipalities, policy, and society. As an outcome of the workshop discussions and building on existing evidence, we highlight three main needs for future science and policy agendas when dealing with nature-based solutions: (i) produce stronger evidence on nature-based solutions for climate change adaptation and mitigation and raise awareness by increasing implementation; (ii) adapt for governance challenges in implementing nature-based solutions by using reflexive approaches, which implies bringing together new networks of society, nature-based solution ambassadors, and practitioners; (iii) consider socio-environmental justice and social cohesion when implementing nature-based solutions by using integrated governance approaches that take into account an integrative and transdisciplinary participation of diverse actors. Taking these needs into account, nature-based solutions can serve as climate mitigation and adaptation tools that produce additional cobenefits for societal well-being, thereby serving as strong investment options for sustainable urban planning.

Abstract: We introduce the dense captioning task, which requires a computer vision system to both localize and describe salient regions in images in natural language. The dense captioning task generalizes object detection when the descriptions consist of a single word, and Image Captioning when one predicted region covers the full image. To address the localization and description task jointly we propose a Fully Convolutional Localization Network (FCLN) architecture that processes an image with a single, efficient forward pass, requires no external regions proposals, and can be trained end-to-end with a single round of optimization. The architecture is composed of a Convolutional Network, a novel dense localization layer, and Recurrent Neural Network language model that generates the label sequences. We evaluate our network on the Visual Genome dataset, which comprises 94,000 images and 4,100,000 region-grounded captions. We observe both speed and accuracy improvements over baselines based on current state of the art approaches in both generation and retrieval settings.

Abstract: We report the INTernational Gamma-ray Astrophysics Laboratory (INTEGRAL) detection of the short gamma-ray burst GRB 170817A (discovered by Fermi-GBM) with a signal-to-noise ratio of 4.6, and, for the first time, its association with the gravitational waves (GWs) from binary neutron star (BNS) merging event GW170817 detected by the LIGO and Virgo observatories. The significance of association between the gamma-ray burst observed by INTEGRAL and GW170817 is 3.2σ, while the association between the Fermi-GBM and INTEGRAL detections is 4.2σ. GRB 170817A was detected by the SPI-ACS instrument about 2 s after the end of the GW event. We measure a fluence of (1.4 ± 0.4 ± 0.6) × 10(−)(7) erg cm(−)(2) (75–2000 keV), where, respectively, the statistical error is given at the 1σ confidence level, and the systematic error corresponds to the uncertainty in the spectral model and instrument response. We also report on the pointed follow-up observations carried out by INTEGRAL, starting 19.5 hr after the event, and lasting for 5.4 days. We provide a stringent upper limit on any electromagnetic signal in a very broad energy range, from 3 keV to 8 MeV, constraining the soft gamma-ray afterglow flux to <7.1 × 10(−)(11) erg cm(−)(2) s(−)(1) (80–300 keV). Exploiting the unique capabilities of INTEGRAL, we constrained the gamma-ray line emission from radioactive decays that are expected to be the principal source of the energy behind a kilonova event following a BNS coalescence. Finally, we put a stringent upper limit on any delayed bursting activity, for example, from a newly formed magnetar.

Abstract: Previous works utilized “smaller-norm-less-important” criterion to prune filters with smaller norm values in a convolutional neural network In this paper, we analyze this norm-based criterion and point out that its effectiveness depends on two requirements that are not always met: (1) the norm deviation of the filters should be large; (2) the minimum norm of the filters should be small To solve this problem, we propose a novel filter pruning method, namely Filter Pruning via Geometric Median (FPGM), to compress the model regardless of those two requirements Unlike previous methods, FPGM compresses CNN models by pruning filters with redundancy, rather than those with“relatively less” importance When applied to two image classification benchmarks, our method validates its usefulness and strengths Notably, on CIFAR-10, FPGM reduces more than 52% FLOPs on ResNet-110 with even 269% relative accuracy improvement Moreover, on ILSVRC-2012, FPGM reduces more than 42% FLOPs on ResNet-101 without top-5 accuracy drop, which has advanced the state-of-the-art Code is publicly available on GitHub: https://githubcom/he-y/filter-pruning-geometric-median

Abstract: Importance The degree to which children and adolescents are infected by and transmit severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is unclear. The role of children and adolescents in transmission of SARS-CoV-2 is dependent on susceptibility, symptoms, viral load, social contact patterns, and behavior. Objective To systematically review the susceptibility to and transmission of SARS-CoV-2 among children and adolescents compared with adults. Data Sources PubMed and medRxiv were searched from database inception to July 28, 2020, and a total of 13 926 studies were identified, with additional studies identified through hand searching of cited references and professional contacts. Study Selection Studies that provided data on the prevalence of SARS-CoV-2 in children and adolescents (younger than 20 years) compared with adults (20 years and older) derived from contact tracing or population screening were included. Single-household studies were excluded. Data Extraction and Synthesis PRISMA guidelines for abstracting data were followed, which was performed independently by 2 reviewers. Quality was assessed using a critical appraisal checklist for prevalence studies. Random-effects meta-analysis was undertaken. Main Outcomes and Measures Secondary infection rate (contact-tracing studies) or prevalence or seroprevalence (population screening studies) among children and adolescents compared with adults. Results A total of 32 studies comprising 41 640 children and adolescents and 268 945 adults met inclusion criteria, including 18 contact-tracing studies and 14 population screening studies. The pooled odds ratio of being an infected contact in children compared with adults was 0.56 (95% CI, 0.37-0.85), with substantial heterogeneity (I2 = 94.6%). Three school-based contact-tracing studies found minimal transmission from child or teacher index cases. Findings from population screening studies were heterogenous and were not suitable for meta-analysis. Most studies were consistent with lower seroprevalence in children compared with adults, although seroprevalence in adolescents appeared similar to adults. Conclusions and Relevance In this meta-analysis, there is preliminary evidence that children and adolescents have lower susceptibility to SARS-CoV-2, with an odds ratio of 0.56 for being an infected contact compared with adults. There is weak evidence that children and adolescents play a lesser role than adults in transmission of SARS-CoV-2 at a population level. This study provides no information on the infectivity of children.

Abstract: We investigate a dilaton gravity model in AdS2 proposed by Almheiri and Polchinski [1] and develop a 1d effective description in terms of a dynamical boundary time with a Schwarzian derivative action. We show that the effective model is equivalent to a 1d version of Liouville theory, and investigate its dynamics and symmetries via a standard canonical framework. We include the coupling to arbitrary conformal matter and analyze the effective action in the presence of possible sources. We compute commutators of local operators at large time separation, and match the result with the time shift due to a gravitational shockwave interaction. We study a black hole evaporation process and comment on the role of entropy in this model.

Abstract: This review presents a comprehensive and systematic study of the field of plant biostimulants and considers the fundamental and innovative principles underlying this technology. The elucidation of the biological basis of biostimulant function is a prerequisite for the development of science-based biostimulant industry and sound regulations governing these compounds. The task of defining the biological basis of biostimulants as a class of compounds, however, is made more complex by the diverse sources of biostimulants present in the market, which include bacteria, fungi, seaweeds, higher plants, animals and humate-containing raw materials, and the wide diversity of industrial processes utilized in their preparation. To distinguish biostimulants from the existing legislative product categories we propose the following definition of a biostimulant as ‘a formulated product of biological origin that improves plant productivity as a consequence of the novel or emergent properties of the complex of constituents, and not as a sole consequence of the presence of known essential plant nutrients, plant growth regulators, or plant protective compounds’. The definition provided here is important as it emphasizes the principle that biological function can be positively modulated through application of molecules, or mixtures of molecules, for which an explicit mode of action has not been defined. Given the difficulty in determining a ‘mode of action’ for a biostimulant, and recognizing the need for the market in biostimulants to attain legitimacy, we suggest that the focus of biostimulant research and validation should be upon proof of efficacy and safety and the determination of a broad mechanism of action, without a requirement for the determination of a specific mode of action. While there is a clear commercial imperative to rationalize biostimulants as a discrete class of products, there is also a compelling biological case for the science-based development of, and experimentation with biostimulants in the expectation that this may lead to the identification of novel biological molecules and phenomenon, pathways and processes, that would not have been discovered if the category of biostimulants did not exist, or was not considered legitimate.

Abstract: Three-dimensional (3D) printing has numerous applications and has gained much interest in the medical world. The constantly improving quality of 3D-printing applications has contributed to their increased use on patients. This paper summarizes the literature on surgical 3D-printing applications used on patients, with a focus on reported clinical and economic outcomes. Three major literature databases were screened for case series (more than three cases described in the same study) and trials of surgical applications of 3D printing in humans. 227 surgical papers were analyzed and summarized using an evidence table. The papers described the use of 3D printing for surgical guides, anatomical models, and custom implants. 3D printing is used in multiple surgical domains, such as orthopedics, maxillofacial surgery, cranial surgery, and spinal surgery. In general, the advantages of 3D-printed parts are said to include reduced surgical time, improved medical outcome, and decreased radiation exposure. The costs of printing and additional scans generally increase the overall cost of the procedure. 3D printing is well integrated in surgical practice and research. Applications vary from anatomical models mainly intended for surgical planning to surgical guides and implants. Our research suggests that there are several advantages to 3D-printed applications, but that further research is needed to determine whether the increased intervention costs can be balanced with the observable advantages of this new technology. There is a need for a formal cost–effectiveness analysis.

Abstract: Tumor metastasis is a major contributor to the death of cancer patients. It is driven not only by the intrinsic alterations in tumor cells, but also by the implicated cross-talk between cancer cells and their altered microenvironment components. Tumor-associated macrophages (TAMs) are the key cells that create an immunosuppressive tumor microenvironment (TME) by producing cytokines, chemokines, growth factors, and triggering the inhibitory immune checkpoint proteins release in T cells. In doing so, TAMs exhibit important functions in facilitating a metastatic cascade of cancer cells and, meanwhile, provide multiple targets of certain checkpoint blockade immunotherapies for opposing tumor progression. In this article, we summarize the regulating networks of TAM polarization and the mechanisms underlying TAM-facilitated metastasis. Based on the overview of current experimental evidence dissecting the critical roles of TAMs in tumor metastasis, we discuss and prospect the potential applications of TAM-focused therapeutic strategies in clinical cancer treatment at present and in the future.

Abstract: We propose a method that can generate an unambiguous description (known as a referring expression) of a specific object or region in an image, and which can also comprehend or interpret such an expression to infer which object is being described. We show that our method outperforms previous methods that generate descriptions of objects without taking into account other potentially ambiguous objects in the scene. Our model is inspired by recent successes of deep learning methods for image captioning, but while image captioning is difficult to evaluate, our task allows for easy objective evaluation. We also present a new large-scale dataset for referring expressions, based on MSCOCO. We have released the dataset and a toolbox for visualization and evaluation, see https://github.com/ mjhucla/Google_Refexp_toolbox.

Abstract: PURPOSE Mixed methods research is becoming an important methodology to investigate complex health-related topics, yet the meaningful integration of qualitative and quantitative data remains elusive and needs further development. A promising innovation to facilitate integration is the use of visual joint displays that bring data together visually to draw out new insights. The purpose of this study was to identify exemplar joint displays by analyzing the various types of joint displays being used in published articles. METHODS We searched for empirical articles that included joint displays in 3 journals that publish state-of-the-art mixed methods research. We analyzed each of 19 identified joint displays to extract the type of display, mixed methods design, purpose, rationale, qualitative and quantitative data sources, integration approaches, and analytic strategies. Our analysis focused on what each display communicated and its representation of mixed methods analysis. RESULTS The most prevalent types of joint displays were statistics-by-themes and side-by-side comparisons. Innovative joint displays connected findings to theoretical frameworks or recommendations. Researchers used joint displays for convergent, explanatory sequential, exploratory sequential, and intervention designs. We identified exemplars for each of these designs by analyzing the inferences gained through using the joint display. Exemplars represented mixed methods integration, presented integrated results, and yielded new insights. CONCLUSIONS Joint displays appear to provide a structure to discuss the integrated analysis and assist both researchers and readers in understanding how mixed methods provides new insights. We encourage researchers to use joint displays to integrate and represent mixed methods analysis and discuss their value.

Abstract: We introduce a large scale MAchine Reading COmprehension dataset, which we name MS MARCO. The dataset comprises of 1,010,916 anonymized questions---sampled from Bing's search query logs---each with a human generated answer and 182,669 completely human rewritten generated answers. In addition, the dataset contains 8,841,823 passages---extracted from 3,563,535 web documents retrieved by Bing---that provide the information necessary for curating the natural language answers. A question in the MS MARCO dataset may have multiple answers or no answers at all. Using this dataset, we propose three different tasks with varying levels of difficulty: (i) predict if a question is answerable given a set of context passages, and extract and synthesize the answer as a human would (ii) generate a well-formed answer (if possible) based on the context passages that can be understood with the question and passage context, and finally (iii) rank a set of retrieved passages given a question. The size of the dataset and the fact that the questions are derived from real user search queries distinguishes MS MARCO from other well-known publicly available datasets for machine reading comprehension and question-answering. We believe that the scale and the real-world nature of this dataset makes it attractive for benchmarking machine reading comprehension and question-answering models.

Abstract: There is compelling evidence that more diverse ecosystems deliver greater benefits to people, and these ecosystem services have become a key argument for biodiversity conservation. However, it is unclear how much biodiversity is needed to deliver ecosystem services in a cost-effective way. Here we show that, while the contribution of wild bees to crop production is significant, service delivery is restricted to a limited subset of all known bee species. Across crops, years and biogeographical regions, crop-visiting wild bee communities are dominated by a small number of common species, and threatened species are rarely observed on crops. Dominant crop pollinators persist under agricultural expansion and many are easily enhanced by simple conservation measures, suggesting that cost-effective management strategies to promote crop pollination should target a different set of species than management strategies to promote threatened bees. Conserving the biological diversity of bees therefore requires more than just ecosystem-service-based arguments.

Abstract: Today in advanced capitalist societies there is a rift between radical theory and radical politics, caused by the twin crises of socialism and Marxism. Socialism as model and Marxism as a theoretical approach are unattractive because of the authoritarianism of existing socialist societies, because Marxism's designated revolutionary agent - the working class - does not act as a class, and because many of today's radical issues - nuclear war, ecology, feminism, gay liberation - have not been compellingly addressed within the socialist tradition. Neither socialism nor Marxism offers an integrative vision for many activists in today's new social movements. Marxism creatively informs work within various academic disciplines, but the Marxist notion that there are overarching, total explanations of social reality is considered dubious, if considered at all. The current contenders among general, transdisciplinary outlooks, such as deconstruction, stress the fragmentary nature of reality, indeterminacy, and the lack of a unitary agency for change. The various post-Marxisms - mainly filiations out of the Age (!) of Structuralism - have been debilitating for radical politics. Although there are important contributions in the "linguistic turn," emphasizing non-economic motivations, the fluidity of social contexts, the power of language, and the rhetorical construction of social movements and cultural trends, on the whole its message has been that materialist determinism has to be replaced with de-ontologized indeterminacy. Deconstruction might mean critical engagement by which wholes are broken apart so their vibrant elements can be reappropriated in changed frameworks. But in casting aside Marxist answers too many of its critics also set aside the questions these answers were meant to address.

Abstract: Numerous important problems can be framed as learning from graph data. We propose a framework for learning convolutional neural networks for arbitrary graphs. These graphs may be undirected, directed, and with both discrete and continuous node and edge attributes. Analogous to image-based convolutional networks that operate on locally connected regions of the input, we present a general approach to extracting locally connected regions from graphs. Using established benchmark data sets, we demonstrate that the learned feature representations are competitive with state of the art graph kernels and that their computation is highly efficient.

Abstract: Importance The data reported herein were accepted for assessment by the US Food and Drug Administration for Biologics License Application under priority review to establish the clinical benefit of durvalumab as second-line therapy for locally advanced or metastatic urothelial carcinoma (UC), resulting in its recent US approval. Objective To report a planned update of the safety and efficacy of durvalumab in patients with locally advanced/metastatic UC. Design, Setting, and Participants This is an ongoing phase 1/2 open-label study of 191 adult patients with histologically or cytologically confirmed locally advanced/metastatic UC whose disease had progressed on, were ineligible for, or refused prior chemotherapy from 60 sites in 9 countries as reported herein. Intervention Patients were administered durvalumab intravenous infusion, 10 mg/kg every 2 weeks, for up to 12 months or until progression, starting another anticancer therapy, or unacceptable toxic effects. Main Outcomes and Measures Primary end points were safety and confirmed objective response rate (ORR) per blinded independent central review (Response Evaluation Criteria In Solid Tumors [RECIST], version 1.1). Results A total of 191 patients with UC had received treatment. As of October 24, 2016 (90-day update), the median follow-up was 5.78 months (range, 0.4-25.9 months). The median age of patients was 67.0 years and most were male (136 [71.2%]) and white (123 [71.1%]). All patients had stage 4 disease, and 190 (99.5%) had prior anticancer therapy (182 [95.3%] postplatinum). The ORR was 17.8% (34 of 191; 95% CI, 12.7%-24.0%), including 7 complete responses. Responses were early (median time to response, 1.41 months), durable (median duration of response not reached), and observed regardless of programmed cell death ligand-1 (PD-L1) expression (ORR, 27.6% [n = 27; 95% CI, 19.0%-37.5%] and 5.1% [n = 4; 95% CI, 1.4%-12.5%] in patients with high and low or negative expression of PD-L1, respectively). Median progression-free survival and overall survival were 1.5 months (95% CI, 1.4-1.9 months) and 18.2 months (95% CI, 8.1 months to not estimable), respectively; the 1-year overall survival rate was 55% (95% CI, 44%-65%), as estimated by Kaplan-Meier method. Grade 3/4 treatment-related adverse events (AEs) occurred in 13 patients (6.8%); grade 3/4 immune-mediated AEs occurred in 4 patients (2.1%); and treatment-related AEs led to discontinuation of 3 patients (1.6%), 2 of whom had immune-mediated AEs that led to death (autoimmune hepatitis and pneumonitis). Conclusions and Relevance Durvalumab, 10 mg/kg every 2 weeks, demonstrates favorable clinical activity and an encouraging and manageable safety profile in patients with locally advanced/metastatic UC. Trial Registration clinicaltrials.gov Identifier:NCT01693562

Abstract: Motivated by vision-based reinforcement learning (RL) problems, in particular Atari games from the recent benchmark Aracade Learning Environment (ALE), we consider spatio-temporal prediction problems where future (image-)frames are dependent on control variables or actions as well as previous frames. While not composed of natural scenes, frames in Atari games are high-dimensional in size, can involve tens of objects with one or more objects being controlled by the actions directly and many other objects being influenced indirectly, can involve entry and departure of objects, and can involve deep partial observability. We propose and evaluate two deep neural network architectures that consist of encoding, action-conditional transformation, and decoding layers based on convolutional neural networks and recurrent neural networks. Experimental results show that the proposed architectures are able to generate visually-realistic frames that are also useful for control over approximately 100-step action-conditional futures in some games. To the best of our knowledge, this paper is the first to make and evaluate long-term predictions on high-dimensional video conditioned by control inputs.

Abstract: Antibody-Mediated Encephalitis The category of autoimmune encephalitides constitutes disorders with relatively distinct characteristics such as psychosis, seizures, abnormal movements, coma, and dysautonomia. Specific autoantibodies can be identified, and the disorders can be successfully treated.

Abstract: CD4+ T cells are central mediators of autoimmune pathology; however, defining their key effector functions in specific autoimmune diseases remains challenging. Pathogenic CD4+ T cells within affected tissues may be identified by expression of markers of recent activation. Here we use mass cytometry to analyse activated T cells in joint tissue from patients with rheumatoid arthritis, a chronic immune-mediated arthritis that affects up to 1% of the population. This approach revealed a markedly expanded population of PD-1hiCXCR5-CD4+ T cells in synovium of patients with rheumatoid arthritis. However, these cells are not exhausted, despite high PD-1 expression. Rather, using multidimensional cytometry, transcriptomics, and functional assays, we define a population of PD-1hiCXCR5- 'peripheral helper' T (TPH) cells that express factors enabling B-cell help, including IL-21, CXCL13, ICOS, and MAF. Like PD-1hiCXCR5+ T follicular helper cells, TPH cells induce plasma cell differentiation in vitro through IL-21 secretion and SLAMF5 interaction (refs 3, 4). However, global transcriptomics highlight differences between TPH cells and T follicular helper cells, including altered expression of BCL6 and BLIMP1 and unique expression of chemokine receptors that direct migration to inflamed sites, such as CCR2, CX3CR1, and CCR5, in TPH cells. TPH cells appear to be uniquely poised to promote B-cell responses and antibody production within pathologically inflamed non-lymphoid tissues.

Abstract: PurposeTo investigate the safety and efficacy of durvalumab, a human monoclonal antibody that binds programmed cell death ligand-1 (PD-L1), and the role of PD-L1 expression on clinical response in patients with advanced urothelial bladder cancer (UBC).MethodsA phase 1/2 multicenter, open-label study is being conducted in patients with inoperable or metastatic solid tumors. We report here the results from the UBC expansion cohort. Durvalumab (MEDI4736, 10 mg/kg every 2 weeks) was administered intravenously for up to 12 months. The primary end point was safety, and objective response rate (ORR, confirmed) was a key secondary end point. An exploratory analysis of pretreatment tumor biopsies led to defining PD-L1–positive as ≥ 25% of tumor cells or tumor-infiltrating immune cells expressing membrane PD-L1.ResultsA total of 61 patients (40 PD-L1–positive, 21 PD-L1–negative), 93.4% of whom received one or more prior therapies for advanced disease, were treated (median duration of follow-up, 4.3 months). The mos...

Abstract: Neural networks are vulnerable to adversarial examples, which poses a threat to their application in security sensitive systems. We propose high-level representation guided denoiser (HGD) as a defense for image classification. Standard denoiser suffers from the error amplification effect, in which small residual adversarial noise is progressively amplified and leads to wrong classifications. HGD overcomes this problem by using a loss function defined as the difference between the target model's outputs activated by the clean image and denoised image. Compared with ensemble adversarial training which is the state-of-the-art defending method on large images, HGD has three advantages. First, with HGD as a defense, the target model is more robust to either white-box or black-box adversarial attacks. Second, HGD can be trained on a small subset of the images and generalizes well to other images and unseen classes. Third, HGD can be transferred to defend models other than the one guiding it. In NIPS competition on defense against adversarial attacks, our HGD solution won the first place and outperformed other models by a large margin.1