Abstract: Earlier detection is key to reducing cancer deaths. Here, we describe a blood test that can detect eight common cancer types through assessment of the levels of circulating proteins and mutations in cell-free DNA. We applied this test, called CancerSEEK, to 1005 patients with nonmetastatic, clinically detected cancers of the ovary, liver, stomach, pancreas, esophagus, colorectum, lung, or breast. CancerSEEK tests were positive in a median of 70% of the eight cancer types. The sensitivities ranged from 69 to 98% for the detection of five cancer types (ovary, liver, stomach, pancreas, and esophagus) for which there are no screening tests available for average-risk individuals. The specificity of CancerSEEK was greater than 99%: only 7 of 812 healthy controls scored positive. In addition, CancerSEEK localized the cancer to a small number of anatomic sites in a median of 83% of the patients.

Abstract: Carbon-based supercapacitors can provide high electrical power, but they do not have sufficient energy density to directly compete with batteries. We found that a nitrogen-doped ordered mesoporous few-layer carbon has a capacitance of 855 farads per gram in aqueous electrolytes and can be bipolarly charged or discharged at a fast, carbon-like speed. The improvement mostly stems from robust redox reactions at nitrogen-associated defects that transform inert graphene-like layered carbon into an electrochemically active substance without affecting its electric conductivity. These bipolar aqueous-electrolyte electrochemical cells offer power densities and lifetimes similar to those of carbon-based supercapacitors and can store a specific energy of 41 watt-hours per kilogram (19.5 watt-hours per liter).

Abstract: Understanding entailment and contradiction is fundamental to understanding natural language, and inference about entailment and contradiction is a valuable testing ground for the development of semantic representations. However, machine learning research in this area has been dramatically limited by the lack of large-scale resources. To address this, we introduce the Stanford Natural Language Inference corpus, a new, freely available collection of labeled sentence pairs, written by humans doing a novel grounded task based on image captioning. At 570K pairs, it is two orders of magnitude larger than all other resources of its type. This increase in scale allows lexicalized classifiers to outperform some sophisticated existing entailment models, and it allows a neural network-based model to perform competitively on natural language inference benchmarks for the first time.

Abstract: For the vast majority of stars in the second Gaia data release, reliable distances cannot be obtained by inverting the parallax. A correct inference procedure must instead be used to account for the nonlinearity of the transformation and the asymmetry of the resulting probability distribution. Here we infer distances to essentially all 1.33 billion stars with parallaxes published in the second \gaia\ data release. This is done using a weak distance prior that varies smoothly as a function of Galactic longitude and latitude according to a Galaxy model. The irreducible uncertainty in the distance estimate is characterized by the lower and upper bounds of an asymmetric confidence interval. Although more precise distances can be estimated for a subset of the stars using additional data (such as photometry), our goal is to provide purely geometric distance estimates, independent of assumptions about the physical properties of, or interstellar extinction towards, individual stars. We analyse the characteristics of the catalogue and validate it using clusters. The catalogue can be queried on the Gaia archive using ADQL at this http URL and downloaded from this http URL .

Abstract: A census of the biomass on Earth is key for understanding the structure and dynamics of the biosphere. However, a global, quantitative view of how the biomass of different taxa compare with one another is still lacking. Here, we assemble the overall biomass composition of the biosphere, establishing a census of the ≈550 gigatons of carbon (Gt C) of biomass distributed among all of the kingdoms of life. We find that the kingdoms of life concentrate at different locations on the planet; plants (≈450 Gt C, the dominant kingdom) are primarily terrestrial, whereas animals (≈2 Gt C) are mainly marine, and bacteria (≈70 Gt C) and archaea (≈7 Gt C) are predominantly located in deep subsurface environments. We show that terrestrial biomass is about two orders of magnitude higher than marine biomass and estimate a total of ≈6 Gt C of marine biota, doubling the previous estimated quantity. Our analysis reveals that the global marine biomass pyramid contains more consumers than producers, thus increasing the scope of previous observations on inverse food pyramids. Finally, we highlight that the mass of humans is an order of magnitude higher than that of all wild mammals combined and report the historical impact of humanity on the global biomass of prominent taxa, including mammals, fish, and plants.

Abstract: BackgroundZika virus (ZIKV) has been linked to central nervous system malformations in fetuses. To characterize the spectrum of ZIKV disease in pregnant women and infants, we followed patients in Rio de Janeiro to describe clinical manifestations in mothers and repercussions of acute ZIKV infection in infants. MethodsWe enrolled pregnant women in whom a rash had developed within the previous 5 days and tested blood and urine specimens for ZIKV by reverse-transcriptase–polymerase-chain-reaction assays. We followed women prospectively to obtain data on pregnancy and infant outcomes. ResultsA total of 345 women were enrolled from September 2015 through May 2016; of these, 182 women (53%) tested positive for ZIKV in blood, urine, or both. The timing of acute ZIKV infection ranged from 6 to 39 weeks of gestation. Predominant maternal clinical features included a pruritic descending macular or maculopapular rash, arthralgias, conjunctival injection, and headache; 27% had fever (short-term and low-grade). By Jul...

Abstract: Social scientists and policy analysts have long expressed concern about the extent of intergenerational income mobility in the United States, but remarkably little empirical evidence is available. The few existing estimates of the intergenerational correlation in income have been biased downward by measurement error, unrepresentative samples, or both. New estimates based on intergenerational data from the Panel Study of Income Dynamics imply that the intergenerational correlation in long-run income is at least 0.4, indicating dramatically less mobility than suggested by earlier research. Copyright 1992 by American Economic Association.

Abstract: Background: This document provides evidence-based clinical practice guidelines on the management of adult patients with community-acquired pneumonia.Methods: A multidisciplinary panel conducted pra...

Abstract: This chapter is a reprint of Frank P. Ramsey’s seminal paper “Truth and Probability” written in 1926 and first published posthumous in the 1931 The Foundations of Mathematics and other Logical Essays, ed. R.B. Braithwaite, London: Routledge & Kegan Paul Ltd.

Abstract: Deep reinforcement learning is poised to revolutionise the field of AI and represents a step towards building autonomous systems with a higher level understanding of the visual world. Currently, deep learning is enabling reinforcement learning to scale to problems that were previously intractable, such as learning to play video games directly from pixels. Deep reinforcement learning algorithms are also applied to robotics, allowing control policies for robots to be learned directly from camera inputs in the real world. In this survey, we begin with an introduction to the general field of reinforcement learning, then progress to the main streams of value-based and policy-based methods. Our survey will cover central algorithms in deep reinforcement learning, including the deep $Q$-network, trust region policy optimisation, and asynchronous advantage actor-critic. In parallel, we highlight the unique advantages of deep neural networks, focusing on visual understanding via reinforcement learning. To conclude, we describe several current areas of research within the field.

Abstract: The ability to perform pixel-wise semantic segmentation in real-time is of paramount importance in mobile applications. Recent deep neural networks aimed at this task have the disadvantage of requiring a large number of floating point operations and have long run-times that hinder their usability. In this paper, we propose a novel deep neural network architecture named ENet (efficient neural network), created specifically for tasks requiring low latency operation. ENet is up to 18$\times$ faster, requires 75$\times$ less FLOPs, has 79$\times$ less parameters, and provides similar or better accuracy to existing models. We have tested it on CamVid, Cityscapes and SUN datasets and report on comparisons with existing state-of-the-art methods, and the trade-offs between accuracy and processing time of a network. We present performance measurements of the proposed architecture on embedded systems and suggest possible software improvements that could make ENet even faster.

Abstract: Summary Background Initial results of the ChemoRadiotherapy for Oesophageal cancer followed by Surgery Study (CROSS) comparing neoadjuvant chemoradiotherapy plus surgery versus surgery alone in patients with squamous cell carcinoma and adenocarcinoma of the oesophagus or oesophagogastric junction showed a significant increase in 5-year overall survival in favour of the neoadjuvant chemoradiotherapy plus surgery group after a median of 45 months' follow-up. In this Article, we report the long-term results after a minimum follow-up of 5 years. Methods Patients with clinically resectable, locally advanced cancer of the oesophagus or oesophagogastric junction (clinical stage T1N1M0 or T2–3N0–1M0, according to the TNM cancer staging system, sixth edition) were randomly assigned in a 1:1 ratio with permuted blocks of four or six to receive either weekly administration of five cycles of neoadjuvant chemoradiotherapy (intravenous carboplatin [AUC 2 mg/mL per min] and intravenous paclitaxel [50 mg/m 2 of body-surface area] for 23 days) with concurrent radiotherapy (41·4 Gy, given in 23 fractions of 1·8 Gy on 5 days per week) followed by surgery, or surgery alone. The primary endpoint was overall survival, analysed by intention-to-treat. No adverse event data were collected beyond those noted in the initial report of the trial. This trial is registered with the Netherlands Trial Register, number NTR487, and has been completed. Findings Between March 30, 2004, and Dec 2, 2008, 368 patients from eight participating centres (five academic centres and three large non-academic teaching hospitals) in the Netherlands were enrolled into this study and randomly assigned to the two treatment groups: 180 to surgery plus neoadjuvant chemoradiotherapy and 188 to surgery alone. Two patients in the neoadjuvant chemoradiotherapy group withdrew consent, so a total of 366 patients were analysed (178 in the neoadjuvant chemoradiotherapy plus surgery group and 188 in the surgery alone group). Of 171 patients who received any neoadjuvant chemoradiotherapy in this group, 162 (95%) were able to complete the entire neoadjuvant chemoradiotherapy regimen. After a median follow-up for surviving patients of 84·1 months (range 61·1–116·8, IQR 70·7–96·6), median overall survival was 48·6 months (95% CI 32·1–65·1) in the neoadjuvant chemoradiotherapy plus surgery group and 24·0 months (14·2–33·7) in the surgery alone group (HR 0·68 [95% CI 0·53–0·88]; log-rank p=0·003). Median overall survival for patients with squamous cell carcinomas was 81·6 months (95% CI 47·2–116·0) in the neoadjuvant chemoradiotherapy plus surgery group and 21·1 months (15·4–26·7) in the surgery alone group (HR 0·48 [95% CI 0·28–0·83]; log-rank p=0·008); for patients with adenocarcinomas, it was 43·2 months (24·9–61·4) in the neoadjuvant chemoradiotherapy plus surgery group and 27·1 months (13·0–41·2) in the surgery alone group (HR 0·73 [95% CI 0·55–0·98]; log-rank p=0·038). Interpretation Long-term follow-up confirms the overall survival benefits for neoadjuvant chemoradiotherapy when added to surgery in patients with resectable oesophageal or oesophagogastric junctional cancer. This improvement is clinically relevant for both squamous cell carcinoma and adenocarcinoma subtypes. Therefore, neoadjuvant chemoradiotherapy according to the CROSS trial followed by surgical resection should be regarded as a standard of care for patients with resectable locally advanced oesophageal or oesophagogastric junctional cancer. Funding Dutch Cancer Foundation (KWF Kankerbestrijding).

Abstract: Neural networks are known to be vulnerable to adversarial examples: inputs that are close to natural inputs but classified incorrectly. In order to better understand the space of adversarial examples, we survey ten recent proposals that are designed for detection and compare their efficacy. We show that all can be defeated by constructing new loss functions. We conclude that adversarial examples are significantly harder to detect than previously appreciated, and the properties believed to be intrinsic to adversarial examples are in fact not. Finally, we propose several simple guidelines for evaluating future proposed defenses.

Abstract: Recent research has revealed that the output of deep neural networks (DNNs) can be easily altered by adding relatively small perturbations to the input vector. In this paper, we analyze an attack in an extremely limited scenario where only one pixel can be modified. For that we propose a novel method for generating one-pixel adversarial perturbations based on differential evolution (DE). It requires less adversarial information (a black-box attack) and can fool more types of networks due to the inherent features of DE. The results show that 67.97% of the natural images in Kaggle CIFAR-10 test dataset and 16.04% of the ImageNet (ILSVRC 2012) test images can be perturbed to at least one target class by modifying just one pixel with 74.03% and 22.91% confidence on average. We also show the same vulnerability on the original CIFAR-10 dataset. Thus, the proposed attack explores a different take on adversarial machine learning in an extreme limited scenario, showing that current DNNs are also vulnerable to such low dimension attacks. Besides, we also illustrate an important application of DE (or broadly speaking, evolutionary computation) in the domain of adversarial machine learning: creating tools that can effectively generate low-cost adversarial attacks against neural networks for evaluating robustness.

Abstract: The transition within business from a linear to a circular economy brings with it a range of practical challenges for companies. The following question is addressed: What are the product design and business model strategies for companies that want to move to a circular economy model? This paper develops a framework of strategies to guide designers and business strategists in the move from a linear to a circular economy. Building on Stahel, the terminology of slowing, closing, and narrowing resource loops is introduced. A list of product design strategies, business model strategies, and examples for key decision-makers in businesses is introduced, to facilitate the move to a circular economy. This framework also opens up a future research agenda for the circular economy.

Abstract: Type I interferons (IFNs) have diverse effects on innate and adaptive immune cells during infection with viruses, bacteria, parasites and fungi, directly and/or indirectly through the induction of other mediators. Type I IFNs are important for host defence against viruses. However, recently, they have been shown to cause immunopathology in some acute viral infections, such as influenza virus infection. Conversely, they can lead to immunosuppression during chronic viral infections, such as lymphocytic choriomeningitis virus infection. During bacterial infections, low levels of type I IFNs may be required at an early stage, to initiate cell-mediated immune responses. High concentrations of type I IFNs may block B cell responses or lead to the production of immunosuppressive molecules, and such concentrations also reduce the responsiveness of macrophages to activation by IFNγ, as has been shown for infections with Listeria monocytogenes and Mycobacterium tuberculosis. Recent studies in experimental models of tuberculosis have demonstrated that prostaglandin E2 and interleukin-1 inhibit type I IFN expression and its downstream effects, demonstrating that a cross-regulatory network of cytokines operates during infectious diseases to provide protection with minimum damage to the host.

Abstract: Importance Understanding skin cancer incidence is critical for planning prevention and treatment strategies and allocating medical resources. However, owing to lack of national reporting and previously nonspecific diagnosis classification, accurate measurement of the US incidence of nonmelanoma skin cancer (NMSC) has been difficult. Objective To estimate the incidence of NMSC (keratinocyte carcinomas) in the US population in 2012 and the incidence of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) in the 2012 Medicare fee-for-service population. Design, Setting, and Participants This study analyzes US government administrative data including the Centers for Medicare & Medicaid Services Physicians Claims databases to calculate totals of skin cancer procedures performed for Medicare beneficiaries from 2006 through 2012 and related parameters. The population-based National Ambulatory Medical Care Survey database was used to estimate NMSC-related office visits for 2012. We combined these analyses to estimate totals of new skin cancer diagnoses and affected individuals in the overall US population. Main Outcomes and Measures Incidence of NMSC in the US population in 2012 and BCC and SCC in the 2012 Medicare fee-for-service population. Results The total number of procedures for skin cancer in the Medicare fee-for-service population increased by 13% from 2 048 517 in 2006 to 2 321 058 in 2012. The age-adjusted skin cancer procedure rate per 100 000 beneficiaries increased from 6075 in 2006 to 7320 in 2012. The number of procedures in Medicare beneficiaries specific for NMSC increased by 14% from 1 918 340 in 2006 to 2 191 100 in 2012. The number of persons with at least 1 procedure for NMSC increased by 14% (from 1 177 618 to 1 336 800) from 2006 through 2012. In the 2012 Medicare fee-for-service population, the age-adjusted procedure rate for BCC and SCC were 3280 and 3278 per 100 000 beneficiaries, respectively. The ratio of BCC to SCC treated in Medicare beneficiaries was 1.0. We estimate the total number of NMSCs in the US population in 2012 at 5 434 193 and the total number of persons in the United States treated for NMSC at 3 315 554. Conclusions and Relevance This study is a thorough nationwide estimate of the incidence of NMSC and provides evidence of continued increases in numbers of skin cancer diagnoses and affected patients in the United States. This study also demonstrates equal incidence rates for BCC and SCC in the Medicare population.

Abstract: Foreword: Bill and Melinda Gates Preface: Peter Singer Acknowledgments Famine, Affluence, and Morality The Singer Solution to World Poverty What Should a Billionaire Give - and What Should You? Index

Abstract: Recurrent Neural Networks can be trained to produce sequences of tokens given some input, as exemplified by recent results in machine translation and image captioning. The current approach to training them consists of maximizing the likelihood of each token in the sequence given the current (recurrent) state and the previous token. At inference, the unknown previous token is then replaced by a token generated by the model itself. This discrepancy between training and inference can yield errors that can accumulate quickly along the generated sequence. We propose a curriculum learning strategy to gently change the training process from a fully guided scheme using the true previous token, towards a less guided scheme which mostly uses the generated token instead. Experiments on several sequence prediction tasks show that this approach yields significant improvements. Moreover, it was used succesfully in our winning entry to the MSCOCO image captioning challenge, 2015.

Abstract: We introduce a new type of deep contextualized word representation that models both (1) complex characteristics of word use (e.g., syntax and semantics), and (2) how these uses vary across linguistic contexts (i.e., to model polysemy). Our word vectors are learned functions of the internal states of a deep bidirectional language model (biLM), which is pre-trained on a large text corpus. We show that these representations can be easily added to existing models and significantly improve the state of the art across six challenging NLP problems, including question answering, textual entailment and sentiment analysis. We also present an analysis showing that exposing the deep internals of the pre-trained network is crucial, allowing downstream models to mix different types of semi-supervision signals.

Abstract: We analysed whole-genome sequences of 560 breast cancers to advance understanding of the driver mutations conferring clonal advantage and the mutational processes generating somatic mutations. We found that 93 protein-coding cancer genes carried probable driver mutations. Some non-coding regions exhibited high mutation frequencies, but most have distinctive structural features probably causing elevated mutation rates and do not contain driver mutations. Mutational signature analysis was extended to genome rearrangements and revealed twelve base substitution and six rearrangement signatures. Three rearrangement signatures, characterized by tandem duplications or deletions, appear associated with defective homologous-recombination-based DNA repair: one with deficient BRCA1 function, another with deficient BRCA1 or BRCA2 function, the cause of the third is unknown. This analysis of all classes of somatic mutation across exons, introns and intergenic regions highlights the repertoire of cancer genes and mutational processes operating, and progresses towards a comprehensive account of the somatic genetic basis of breast cancer.

Abstract: We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2. To achieve this result, we first train an EfficientNet model on labeled ImageNet images and use it as a teacher to generate pseudo labels on 300M unlabeled images. We then train a larger EfficientNet as a student model on the combination of labeled and pseudo labeled images. We iterate this process by putting back the student as the teacher. During the generation of the pseudo labels, the teacher is not noised so that the pseudo labels are as accurate as possible. However, during the learning of the student, we inject noise such as dropout, stochastic depth and data augmentation via RandAugment to the student so that the student generalizes better than the teacher.

Abstract: Neural networks are both computationally intensive and memory intensive, making them difficult to deploy on embedded systems. Also, conventional networks fix the architecture before training starts; as a result, training cannot improve the architecture. To address these limitations, we describe a method to reduce the storage and computation required by neural networks by an order of magnitude without affecting their accuracy by learning only the important connections. Our method prunes redundant connections using a three-step method. First, we train the network to learn which connections are important. Next, we prune the unimportant connections. Finally, we retrain the network to fine tune the weights of the remaining connections. On the ImageNet dataset, our method reduced the number of parameters of AlexNet by a factor of 9x, from 61 million to 6.7 million, without incurring accuracy loss. Similar experiments with VGG-16 found that the number of parameters can be reduced by 13x, from 138 million to 10.3 million, again with no loss of accuracy.

Abstract: Recent years have produced great advances in training large, deep neural networks (DNNs), including notable successes in training convolutional neural networks (convnets) to recognize natural images. However, our understanding of how these models work, especially what computations they perform at intermediate layers, has lagged behind. Progress in the field will be further accelerated by the development of better tools for visualizing and interpreting neural nets. We introduce two such tools here. The first is a tool that visualizes the activations produced on each layer of a trained convnet as it processes an image or video (e.g. a live webcam stream). We have found that looking at live activations that change in response to user input helps build valuable intuitions about how convnets work. The second tool enables visualizing features at each layer of a DNN via regularized optimization in image space. Because previous versions of this idea produced less recognizable images, here we introduce several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations. Both tools are open source and work on a pre-trained convnet with minimal setup.

Abstract: The coronavirus disease (COVID-19) has been identified as the cause of an outbreak of respiratory illness in Wuhan, Hubei Province, China beginning in December 2019. As of 31 January 2020, this epidemic had spread to 19 countries with 11 791 confirmed cases, including 213 deaths. The World Health Organization has declared it a Public Health Emergency of International Concern. A scoping review was conducted following the methodological framework suggested by Arksey and O’Malley. In this scoping review, 65 research articles published before 31 January 2020 were analyzed and discussed to better understand the epidemiology, causes, clinical diagnosis, prevention and control of this virus. The research domains, dates of publication, journal language, authors’ affiliations, and methodological characteristics were included in the analysis. All the findings and statements in this review regarding the outbreak are based on published information as listed in the references. Most of the publications were written using the English language (89.2%). The largest proportion of published articles were related to causes (38.5%) and a majority (67.7%) were published by Chinese scholars. Research articles initially focused on causes, but over time there was an increase of the articles related to prevention and control. Studies thus far have shown that the virus’ origination is in connection to a seafood market in Wuhan, but specific animal associations have not been confirmed. Reported symptoms include fever, cough, fatigue, pneumonia, headache, diarrhea, hemoptysis, and dyspnea. Preventive measures such as masks, hand hygiene practices, avoidance of public contact, case detection, contact tracing, and quarantines have been discussed as ways to reduce transmission. To date, no specific antiviral treatment has proven effective; hence, infected people primarily rely on symptomatic treatment and supportive care. There has been a rapid surge in research in response to the outbreak of COVID-19. During this early period, published research primarily explored the epidemiology, causes, clinical manifestation and diagnosis, as well as prevention and control of the novel coronavirus. Although these studies are relevant to control the current public emergency, more high-quality research is needed to provide valid and reliable ways to manage this kind of public health emergency in both the short- and long-term.

Abstract: BackgroundProstate cancer is a heterogeneous disease, but current treatments are not based on molecular stratification. We hypothesized that metastatic, castration-resistant prostate cancers with DNA-repair defects would respond to poly(adenosine diphosphate [ADP]–ribose) polymerase (PARP) inhibition with olaparib. MethodsWe conducted a phase 2 trial in which patients with metastatic, castration-resistant prostate cancer were treated with olaparib tablets at a dose of 400 mg twice a day. The primary end point was the response rate, defined either as an objective response according to Response Evaluation Criteria in Solid Tumors, version 1.1, or as a reduction of at least 50% in the prostate-specific antigen level or a confirmed reduction in the circulating tumor-cell count from 5 or more cells per 7.5 ml of blood to less than 5 cells per 7.5 ml. Targeted next-generation sequencing, exome and transcriptome analysis, and digital polymerase-chain-reaction testing were performed on samples from mandated tumor...