Showing papers by "Université de Montréal published in 2017"
••
Theo Vos1, Amanuel Alemu Abajobir, Kalkidan Hassen Abate2, Cristiana Abbafati3 +775 more•Institutions (305)
TL;DR: The Global Burden of Diseases, Injuries, and Risk Factors Study 2016 (GBD 2016) provides a comprehensive assessment of prevalence, incidence, and years lived with disability (YLDs) for 328 causes in 195 countries and territories from 1990 to 2016.
10,401 citations
••
Mohsen Naghavi1, Amanuel Alemu Abajobir2, Cristiana Abbafati3, Kaja Abbas4 +598 more•Institutions (31)
TL;DR: The Global Burden of Disease 2016 Study (GBD 2016) provides a comprehensive assessment of cause-specific mortality for 264 causes in 195 locations from 1980 to 2016 as discussed by the authors, which includes evaluation of the expected epidemiological transition with changes in development and where local patterns deviate from these trends.
3,228 citations
••
TL;DR: A fast and accurate fully automatic method for brain tumor segmentation which is competitive both in terms of accuracy and speed compared to the state of the art, and introduces a novel cascaded architecture that allows the system to more accurately model local label dependencies.
2,538 citations
••
Johns Hopkins University1, Leipzig University2, Korea University3, Yale University4, West Virginia University5, University of Barcelona6, St George's, University of London7, Indiana University8, National Yang-Ming University9, Cleveland Clinic10, Aarhus University11, University at Buffalo12, Imperial College London13, Primary Children's Hospital14, Erasmus University Rotterdam15, Yeshiva University16, Ghent University17, Baylor University18, Virginia Commonwealth University19, Harvard University20, Federal University of São Paulo21, University of California, San Francisco22, Beaumont Hospital23, Boston University24, University of Oklahoma25, University of Michigan26, Carlos III Health Institute27, University of Melbourne28, Saint Louis University29, Université de Montréal30, University of Pennsylvania31, McGill University32, Mayo Clinic33, Lahey Hospital & Medical Center34, Royal Adelaide Hospital35, University of Milan36, University of Toronto37, Loyola University Chicago38, Jikei University School of Medicine39
TL;DR: This 2017 Consensus Statement is to provide a state-of-the-art review of the field of catheter and surgical ablation of AF and to report the findings of a writing group, convened by these five international societies.
1,626 citations
••
TL;DR: The InTBIR Participants and Investigators have provided informed consent for the study to take place in Poland.
Abstract: Additional co-authors: Endre Czeiter, Marek Czosnyka, Ramon Diaz-Arrastia, Jens P Dreier, Ann-Christine Duhaime, Ari Ercole, Thomas A van Essen, Valery L Feigin, Guoyi Gao, Joseph Giacino, Laura E Gonzalez-Lara, Russell L Gruen, Deepak Gupta, Jed A Hartings, Sean Hill, Ji-yao Jiang, Naomi Ketharanathan, Erwin J O Kompanje, Linda Lanyon, Steven Laureys, Fiona Lecky, Harvey Levin, Hester F Lingsma, Marc Maegele, Marek Majdan, Geoffrey Manley, Jill Marsteller, Luciana Mascia, Charles McFadyen, Stefania Mondello, Virginia Newcombe, Aarno Palotie, Paul M Parizel, Wilco Peul, James Piercy, Suzanne Polinder, Louis Puybasset, Todd E Rasmussen, Rolf Rossaint, Peter Smielewski, Jeannette Soderberg, Simon J Stanworth, Murray B Stein, Nicole von Steinbuchel, William Stewart, Ewout W Steyerberg, Nino Stocchetti, Anneliese Synnot, Braden Te Ao, Olli Tenovuo, Alice Theadom, Dick Tibboel, Walter Videtta, Kevin K W Wang, W Huw Williams, Kristine Yaffe for the InTBIR Participants and Investigators
1,354 citations
••
TL;DR: The size of a planet is an observable property directly connected to the physics of its formation and evolution as discussed by the authors, and the size of close-in (P < 100 days) small planets can be divided into two size regimes: R_p < 1.5 R⊕ or smaller with varying amounts of low-density gas that determine their total sizes.
Abstract: The size of a planet is an observable property directly connected to the physics of its formation and evolution. We used precise radius measurements from the California-Kepler Survey to study the size distribution of 2025 Kepler planets in fine detail. We detect a factor of ≥2 deficit in the occurrence rate distribution at 1.5–2.0 R⊕. This gap splits the population of close-in (P < 100 days) small planets into two size regimes: R_p < 1.5 R⊕ and R_p = 2.0-3.0 R⊕, with few planets in between. Planets in these two regimes have nearly the same intrinsic frequency based on occurrence measurements that account for planet detection efficiencies. The paucity of planets between 1.5 and 2.0 R⊕ supports the emerging picture that close-in planets smaller than Neptune are composed of rocky cores measuring 1.5 R⊕ or smaller with varying amounts of low-density gas that determine their total sizes.
1,100 citations
•
06 Aug 2017
TL;DR: The analysis suggests that the notions of effective capacity which are dataset independent are unlikely to explain the generalization performance of deep networks when trained with gradient based methods because training data itself plays an important role in determining the degree of memorization.
Abstract: We examine the role of memorization in deep learning, drawing connections to capacity, generalization, and adversarial robustness. While deep networks are capable of memorizing noise data, our results suggest that they tend to prioritize learning simple patterns first. In our experiments, we expose qualitative differences in gradient-based optimization of deep neural networks (DNNs) on noise vs. real data. We also demonstrate that for appropriately tuned explicit regularization (e.g., dropout) we can degrade DNN training performance on noise datasets without compromising generalization on real data. Our analysis suggests that the notions of effective capacity which are dataset independent are unlikely to explain the generalization performance of deep networks when trained with gradient based methods because training data itself plays an important role in determining the degree of memorization.
1,080 citations
••
13 Jun 2017
TL;DR: This work describes the ongoing collection of the “something-something” database of video prediction tasks whose solutions require a common sense understanding of the depicted situation, and describes the challenges in crowd-sourcing this data at scale.
Abstract: Neural networks trained on datasets such as ImageNet have led to major advances in visual object classification. One obstacle that prevents networks from reasoning more deeply about complex scenes and situations, and from integrating visual knowledge with natural language, like humans do, is their lack of common sense knowledge about the physical world. Videos, unlike still images, contain a wealth of detailed information about the physical world. However, most labelled video datasets represent high-level concepts rather than detailed physical aspects about actions and scenes. In this work, we describe our ongoing collection of the “something-something” database of video prediction tasks whose solutions require a common sense understanding of the depicted situation. The database currently contains more than 100,000 videos across 174 classes, which are defined as caption-templates. We also describe the challenges in crowd-sourcing this data at scale.
1,062 citations
•
TL;DR: Graph Attention Networks (GATs) as discussed by the authors leverage masked self-attentional layers to address the shortcomings of prior methods based on graph convolutions or their approximations.
Abstract: We present graph attention networks (GATs), novel neural network architectures that operate on graph-structured data, leveraging masked self-attentional layers to address the shortcomings of prior methods based on graph convolutions or their approximations. By stacking layers in which nodes are able to attend over their neighborhoods' features, we enable (implicitly) specifying different weights to different nodes in a neighborhood, without requiring any kind of costly matrix operation (such as inversion) or depending on knowing the graph structure upfront. In this way, we address several key challenges of spectral-based graph neural networks simultaneously, and make our model readily applicable to inductive as well as transductive problems. Our GAT models have achieved or matched state-of-the-art results across four established transductive and inductive graph benchmarks: the Cora, Citeseer and Pubmed citation network datasets, as well as a protein-protein interaction dataset (wherein test graphs remain unseen during training).
1,016 citations
••
Kyriaki Michailidou1, Kyriaki Michailidou2, Sara Lindström3, Sara Lindström4 +393 more•Institutions (127)
TL;DR: A genome-wide association study of breast cancer in 122,977 cases and 105,974 controls of European ancestry and 14,068 cases and 13,104 controls of East Asian ancestry finds that heritability of Breast cancer due to all single-nucleotide polymorphisms in regulatory features was 2–5-fold enriched relative to the genome- wide average.
Abstract: Breast cancer risk is influenced by rare coding variants in susceptibility genes, such as BRCA1, and many common, mostly non-coding variants. However, much of the genetic contribution to breast cancer risk remains unknown. Here we report the results of a genome-wide association study of breast cancer in 122,977 cases and 105,974 controls of European ancestry and 14,068 cases and 13,104 controls of East Asian ancestry. We identified 65 new loci that are associated with overall breast cancer risk at P < 5 × 10-8. The majority of credible risk single-nucleotide polymorphisms in these loci fall in distal regulatory elements, and by integrating in silico data to predict target genes in breast cells at each locus, we demonstrate a strong overlap between candidate target genes and somatic driver genes in breast tumours. We also find that heritability of breast cancer due to all single-nucleotide polymorphisms in regulatory features was 2-5-fold enriched relative to the genome-wide average, with strong enrichment for particular transcription factor binding sites. These results provide further insight into genetic susceptibility to breast cancer and will improve the use of genetic risk scores for individualized screening and prevention.
1,014 citations
••
German Cancer Research Center1, Université de Sherbrooke2, University Health Network3, University of Pittsburgh4, IMT Institute for Advanced Studies Lucca5, St. Jude Children's Research Hospital6, University of Toronto7, Zhejiang University of Technology8, Harvard University9, Utrecht University10, Université de Montréal11, National Research Council12, University of Washington13, University of Western Ontario14, École Polytechnique Fédérale de Lausanne15, ETSI16, Siemens17, University of Southern California18, King's College London19, University of Bordeaux20, Centre national de la recherche scientifique21, Copenhagen University Hospital22, University of Hamburg23, University of Basel24
TL;DR: The encouraging finding that most state-of-the-art algorithms produce tractograms containing 90% of the ground truth bundles (to at least some extent) is reported, however, the same tractograms contain many more invalid than valid bundles, and half of these invalid bundles occur systematically across research groups.
Abstract: Tractography based on non-invasive diffusion imaging is central to the study of human brain connectivity. To date, the approach has not been systematically validated in ground truth studies. Based on a simulated human brain data set with ground truth tracts, we organized an open international tractography challenge, which resulted in 96 distinct submissions from 20 research groups. Here, we report the encouraging finding that most state-of-the-art algorithms produce tractograms containing 90% of the ground truth bundles (to at least some extent). However, the same tractograms contain many more invalid than valid bundles, and half of these invalid bundles occur systematically across research groups. Taken together, our results demonstrate and confirm fundamental ambiguities inherent in tract reconstruction based on orientation information alone, which need to be considered when interpreting tractography and connectivity results. Our approach provides a novel framework for estimating reliability of tractography and encourages innovation to address its current limitations.
•
TL;DR: In this paper, a method to train quantized neural networks (QNNs) with extremely low precision (e.g., 1-bit) weights and activations, at run-time is introduced.
Abstract: We introduce a method to train Quantized Neural Networks (QNNs) -- neural networks with extremely low precision (e.g., 1-bit) weights and activations, at run-time. At traintime the quantized weights and activations are used for computing the parameter gradients. During the forward pass, QNNs drastically reduce memory size and accesses, and replace most arithmetic operations with bit-wise operations. As a result, power consumption is expected to be drastically reduced. We trained QNNs over the MNIST, CIFAR-10, SVHN and ImageNet datasets. The resulting QNNs achieve prediction accuracy comparable to their 32-bit counterparts. For example, our quantized version of AlexNet with 1-bit weights and 2-bit activations achieves 51% top-1 accuracy. Moreover, we quantize the parameter gradients to 6-bits as well which enables gradients computation using only bit-wise operation. Quantized recurrent neural networks were tested over the Penn Treebank dataset, and achieved comparable accuracy as their 32-bit counterparts using only 4-bits. Last but not least, we programmed a binary matrix multiplication GPU kernel with which it is possible to run our MNIST QNN 7 times faster than with an unoptimized GPU kernel, without suffering any loss in classification accuracy. The QNN code is available online.
••
TL;DR: To facilitate network-based approaches to drug discovery, BioGRID now incorporates 27 501 chemical–protein interactions for human drug targets, as drawn from the DrugBank database.
Abstract: The Biological General Repository for Interaction Datasets (BioGRID: https://thebiogrid.org) is an open access database dedicated to the annotation and archival of protein, genetic and chemical interactions for all major model organism species and humans. As of September 2016 (build 3.4.140), the BioGRID contains 1 072 173 genetic and protein interactions, and 38 559 post-translational modifications, as manually annotated from 48 114 publications. This dataset represents interaction records for 66 model organisms and represents a 30% increase compared to the previous 2015 BioGRID update. BioGRID curates the biomedical literature for major model organism species, including humans, with a recent emphasis on central biological processes and specific human diseases. To facilitate network-based approaches to drug discovery, BioGRID now incorporates 27 501 chemical-protein interactions for human drug targets, as drawn from the DrugBank database. A new dynamic interaction network viewer allows the easy navigation and filtering of all genetic and protein interaction data, as well as for bioactive compounds and their established targets. BioGRID data are directly downloadable without restriction in a variety of standardized formats and are freely distributed through partner model organism databases and meta-databases.
••
TL;DR: Vascular normalization, by restoring proper tumor perfusion and oxygenation, could limit tumor cell invasiveness and improve the effectiveness of anticancer treatments.
Abstract: Tumor blood vessels are a key target for cancer therapeutic management. Tumor cells secrete high levels of pro-angiogenic factors which contribute to the creation of an abnormal vascular network characterized by disorganized, immature and permeable blood vessels, resulting in poorly perfused tumors. The hypoxic microenvironment created by impaired tumor perfusion can promote the selection of more invasive and aggressive tumor cells and can also impede the tumor-killing action of immune cells. Furthermore, abnormal tumor perfusion also reduces the diffusion of chemotherapeutic drugs and radiotherapy efficiency. To fight against this defective phenotype, the normalization of the tumor vasculature has emerged as a new therapeutic strategy. Vascular normalization, by restoring proper tumor perfusion and oxygenation, could limit tumor cell invasiveness and improve the effectiveness of anticancer treatments. In this review, we investigate the mechanisms involved in tumor angiogenesis and describe strategies used to achieve vascular normalization.
•
TL;DR: This paper proposed a self-attention mechanism and a special regularization term for the model, which achieved a significant performance gain compared to other sentence embedding methods in all of the three tasks.
Abstract: This paper proposes a new model for extracting an interpretable sentence embedding by introducing self-attention. Instead of using a vector, we use a 2-D matrix to represent the embedding, with each row of the matrix attending on a different part of the sentence. We also propose a self-attention mechanism and a special regularization term for the model. As a side effect, the embedding comes with an easy way of visualizing what specific parts of the sentence are encoded into the embedding. We evaluate our model on 3 different tasks: author profiling, sentiment classification, and textual entailment. Results show that our model yields a significant performance gain compared to other sentence embedding methods in all of the 3 tasks.
••
TL;DR: The aim of this review article is to provide an overview of GANs for the signal processing community, drawing on familiar analogies and concepts where possible, and point to remaining challenges in their theory and application.
Abstract: Generative adversarial networks (GANs) provide a way to learn deep representations without extensively annotated training data. They achieve this through deriving backpropagation signals through a competitive process involving a pair of networks. The representations that can be learned by GANs may be used in a variety of applications, including image synthesis, semantic image editing, style transfer, image super-resolution and classification. The aim of this review paper is to provide an overview of GANs for the signal processing community, drawing on familiar analogies and concepts where possible. In addition to identifying different methods for training and constructing GANs, we also point to remaining challenges in their theory and application.
•
09 Mar 2017TL;DR: A new model for extracting an interpretable sentence embedding by introducing self-attention is proposed, which uses a 2-D matrix to represent the embedding, with each row of the matrix attending on a different part of the sentence.
Abstract: This paper proposes a new model for extracting an interpretable sentence embedding by introducing self-attention. Instead of using a vector, we use a 2-D matrix to represent the embedding, with each row of the matrix attending on a different part of the sentence. We also propose a self-attention mechanism and a special regularization term for the model. As a side effect, the embedding comes with an easy way of visualizing what specific parts of the sentence are encoded into the embedding. We evaluate our model on 3 different tasks: author profiling, sentiment classification, and textual entailment. Results show that our model yields a significant performance gain compared to other sentence embedding methods in all of the 3 tasks.
••
University of Oxford1, Wellington Management Company2, University of Barcelona3, University of Melbourne4, University of Amsterdam5, Ghent University Hospital6, Erasmus University Rotterdam7, National Institutes of Health8, Imperial College London9, Université de Montréal10, University of California, San Francisco11, Boston Children's Hospital12, University of Newcastle13, John Hunter Hospital14, Queen's University Belfast15, University of Western Australia16, French Institute of Health and Medical Research17, Université Paris-Saclay18, University of New South Wales19, University of Arizona20, Ludwig Maximilian University of Munich21, University of Pittsburgh22, University of Cape Town23
TL;DR: The only way to make progress in the future is to be much more clear about the meaning of the labels used for asthma and to acknowledge the assumptions associated with them, which are believed to be the most important causes of the stagnation in key clinical outcomes observed in the past 10 years.
••
University of Tehran1, Université de Montréal2, New Mexico State University3, Royal Botanic Gardens4, State University of Feira de Santana5, State University of Campinas6, University of the Western Cape7, Federal University of São Carlos8, University of Melbourne9, Federal University of Bahia10, National Taiwan University11, Australian National University12, Complutense University of Madrid13, National Autonomous University of Mexico14, Cornell University15, Université libre de Bruxelles16, National Museum of Natural History17, University of Oxford18, Sao Paulo State University19, Universidad de Morón20, Federal University of Western Bahia21, Royal Botanic Garden Edinburgh22, University of Reading23, University of Zurich24, Universidade Federal do Rio Grande do Sul25, Kyushu University26, University of South Africa27, Tarbiat Modares University28, Montana State University29, University of Johannesburg30, Pontifical Catholic University of Rio de Janeiro31, University of Angers32, National Science Foundation33, Missouri Botanical Garden34, National University of Rosario35, University of Arizona36, Federal University of Rio Grande do Norte37, Universidade Federal de Goiás38, Empresa Brasileira de Pesquisa Agropecuária39, University of Dundee40, Arizona State University at the Polytechnic campus41, Arizona State University42, University of Cape Town43, New York Botanical Garden44, Naturalis45, Heidelberg University46, Chinese Academy of Sciences47
TL;DR: The classification of the legume family proposed here addresses the long-known non-monophyly of the traditionally recognised subfamily Caesalpinioideae, by recognising six robustly supported monophyletic subfamilies and reflects the phylogenetic structure that is consistently resolved.
Abstract: The classification of the legume family proposed here addresses the long-known non-monophyly of the traditionally recognised subfamily Caesalpinioideae, by recognising six robustly supported monophyletic subfamilies. This new classification uses as its framework the most comprehensive phylogenetic analyses of legumes to date, based on plastid matK gene sequences, and including near-complete sampling of genera (698 of the currently recognised 765 genera) and ca. 20% (3696) of known species. The matK gene region has been the most widely sequenced across the legumes, and in most legume lineages, this gene region is sufficiently variable to yield well-supported clades. This analysis resolves the same major clades as in other phylogenies of whole plastid and nuclear gene sets (with much sparser taxon sampling). Our analysis improves upon previous studies that have used large phylogenies of the Leguminosae for addressing evolutionary questions, because it maximises generic sampling and provides a phylogenetic tree that is based on a fully curated set of sequences that are vouchered and taxonomically validated. The phylogenetic trees obtained and the underlying data are available to browse and download, facilitating subsequent analyses that require evolutionary trees. Here we propose a new community-endorsed classification of the family that reflects the phylogenetic structure that is consistently resolved and recognises six subfamilies in Leguminosae: a recircumscribed Caesalpinioideae DC., Cercidoideae Legume Phylogeny Working Group (stat. nov.), Detarioideae Burmeist., Dialioideae Legume Phylogeny Working Group (stat. nov.), Duparquetioideae Legume Phylogeny Working Group (stat. nov.), and Papilionoideae DC. The traditionally recognised subfamily Mimosoideae is a distinct clade nested within the recircumscribed Caesalpinioideae and is referred to informally as the mimosoid clade pending a forthcoming formal tribal and/or cladebased classification of the new Caesalpinioideae. We provide a key for subfamily identification, descriptions with diagnostic charactertistics for the subfamilies, figures illustrating their floral and fruit diversity, and lists of genera by subfamily. This new classification of Leguminosae represents a consensus view of the international legume systematics community; it invokes both compromise and practicality of use.
••
TL;DR: The key concepts of deep learning for clinical radiologists are reviewed, technical requirements are discussed, emerging applications in clinical radiology are described, and limitations and future directions in this field are outlined.
Abstract: Deep learning is a class of machine learning methods that are gaining success and attracting interest in many domains, including computer vision, speech recognition, natural language processing, and playing games. Deep learning methods produce a mapping from raw inputs to desired outputs (eg, image classes). Unlike traditional machine learning methods, which require hand-engineered feature extraction from inputs, deep learning methods learn these features directly from data. With the advent of large datasets and increased computing power, these methods can produce models with exceptional performance. These models are multilayer artificial neural networks, loosely inspired by biologic neural systems. Weighted connections between nodes (neurons) in the network are iteratively adjusted based on example pairs of inputs and target outputs by back-propagating a corrective error signal through the network. For computer vision tasks, convolutional neural networks (CNNs) have proven to be effective. Recently, several clinical applications of CNNs have been proposed and studied in radiology for classification, detection, and segmentation tasks. This article reviews the key concepts of deep learning for clinical radiologists, discusses technical requirements, describes emerging applications in clinical radiology, and outlines limitations and future directions in this field. Radiologists should become familiar with the principles and potential applications of deep learning in medical imaging. ©RSNA, 2017.
••
TL;DR: In this phase 2, multicenter, double‐blind, placebo‐controlled, multiple‐ascending‐dose trial of inclisiran administered as a subcutaneous injection in patients at high risk for cardiovascular disease who had elevated LDL cholesterol levels, inclisIRan was found to lower PCSK9 and LDL cholesterol Levels among patients athigh cardiovascular risk.
Abstract: BackgroundIn a previous study, a single injection of inclisiran, a chemically synthesized small interfering RNA designed to target PCSK9 messenger RNA, was found to produce sustained reductions in low-density lipoprotein (LDL) cholesterol levels over the course of 84 days in healthy volunteers. MethodsWe conducted a phase 2, multicenter, double-blind, placebo-controlled, multiple-ascending-dose trial of inclisiran administered as a subcutaneous injection in patients at high risk for cardiovascular disease who had elevated LDL cholesterol levels. Patients were randomly assigned to receive a single dose of placebo or 200, 300, or 500 mg of inclisiran or two doses (at days 1 and 90) of placebo or 100, 200, or 300 mg of inclisiran. The primary end point was the change from baseline in LDL cholesterol level at 180 days. Safety data were available through day 210, and data on LDL cholesterol and proprotein convertase subtilisin–kexin type 9 (PCSK9) levels were available through day 240. ResultsA total of 501 pa...
••
TL;DR: AI in medicine, which is the focus of this review, has two main branches: virtual and physical, and the virtual branch includes informatics approaches from deep learning information management to control of health management systems, and active guidance of physicians in their treatment decisions.
Abstract: Artificial Intelligence (AI) is a general term that implies the use of a computer to model intelligent behavior with minimal human intervention. AI is generally accepted as having started with the invention of robots. The term derives from the Czech word robota, meaning biosynthetic machines used as forced labor. In this field, Leonardo Da Vinci's lasting heritage is today's burgeoning use of robotic-assisted surgery, named after him, for complex urologic and gynecologic procedures. Da Vinci's sketchbooks of robots helped set the stage for this innovation. AI, described as the science and engineering of making intelligent machines, was officially born in 1956. The term is applicable to a broad range of items in medicine such as robotics, medical diagnosis, medical statistics, and human biology-up to and including today's "omics". AI in medicine, which is the focus of this review, has two main branches: virtual and physical. The virtual branch includes informatics approaches from deep learning information management to control of health management systems, including electronic health records, and active guidance of physicians in their treatment decisions. The physical branch is best represented by robots used to assist the elderly patient or the attending surgeon. Also embodied in this branch are targeted nanorobots, a unique new drug delivery system. The societal and ethical complexities of these applications require further reflection, proof of their medical utility, economic value, and development of interdisciplinary strategies for their wider application.
•
TL;DR: Feature-wise linear modulation (FiLM) as mentioned in this paper is a general-purpose conditioning method for neural networks, which can influence neural network computation via a simple, feature-wise affine transformation based on conditioning information.
Abstract: We introduce a general-purpose conditioning method for neural networks called FiLM: Feature-wise Linear Modulation. FiLM layers influence neural network computation via a simple, feature-wise affine transformation based on conditioning information. We show that FiLM layers are highly effective for visual reasoning - answering image-related questions which require a multi-step, high-level process - a task which has proven difficult for standard deep learning methods that do not explicitly model reasoning. Specifically, we show on visual reasoning tasks that FiLM layers 1) halve state-of-the-art error for the CLEVR benchmark, 2) modulate features in a coherent manner, 3) are robust to ablations and architectural modifications, and 4) generalize well to challenging, new data from few examples or even zero-shot.
••
TL;DR: A concise summary of known structural and functional features of the human insular cortex with a focus on lesion case studies and recent neuroimaging evidence for considerable functional heterogeneity of this brain region is provided.
Abstract: The insular cortex, or "Island of Reil," is hidden deep within the lateral sulcus of the brain. Subdivisions within the insula have been identified on the basis of cytoarchitectonics, sulcal landmarks, and connectivity. Depending on the parcellation technique used, the insula can be divided into anywhere between 2 and 13 distinct subdivisions. The insula subserves a wide variety of functions in humans ranging from sensory and affective processing to high-level cognition. Here, we provide a concise summary of known structural and functional features of the human insular cortex with a focus on lesion case studies and recent neuroimaging evidence for considerable functional heterogeneity of this brain region.
••
TL;DR: Insight into potential clinical application of adenosinergic and other purinergic‐targeting therapies and forecast how these might develop in combination with other anti‐cancer modalities are provided.
Abstract: Cancers are able to grow by subverting immune suppressive pathways, to prevent the malignant cells as being recognized as dangerous or foreign. This mechanism prevents the cancer from being eliminated by the immune system and allows disease to progress from a very early stage to a lethal state. Immunotherapies are newly developing interventions that modify the patient's immune system to fight cancer, by either directly stimulating rejection-type processes or blocking suppressive pathways. Extracellular adenosine generated by the ectonucleotidases CD39 and CD73 is a newly recognized "immune checkpoint mediator" that interferes with anti-tumor immune responses. In this review, we focus on CD39 and CD73 ectoenzymes and encompass aspects of the biochemistry of these molecules as well as detailing the distribution and function on immune cells. Effects of CD39 and CD73 inhibition in preclinical and clinical studies are discussed. Finally, we provide insights into potential clinical application of adenosinergic and other purinergic-targeting therapies and forecast how these might develop in combination with other anti-cancer modalities.
••
Oregon Health & Science University1, Mayo Clinic2, Johns Hopkins University3, Université Paris-Saclay4, University of Texas MD Anderson Cancer Center5, Monash University6, Kaiser Permanente7, Université de Montréal8, Heidelberg University9, Complutense University of Madrid10, University of Copenhagen11, Bristol-Myers Squibb12, VU University Amsterdam13, Radboud University Nijmegen14
TL;DR: Ipilimumab did not improve OS in patients with metastatic castration-resistant prostate cancer and the observed increases in progression-free survival and prostate-specific antigen response rates suggest antitumor activity in a patient subset.
Abstract: Purpose Ipilimumab increases antitumor T-cell responses by binding to cytotoxic T-lymphocyte antigen 4. We evaluated treatment with ipilimumab in asymptomatic or minimally symptomatic patients with chemotherapy-naive metastatic castration-resistant prostate cancer without visceral metastases. Patients and Methods In this multicenter, double-blind, phase III trial, patients were randomly assigned (2:1) to ipilimumab 10 mg/kg or placebo every 3 weeks for up to four doses. Ipilimumab 10 mg/kg or placebo maintenance therapy was administered to nonprogressing patients every 3 months. The primary end point was overall survival (OS). Results Four hundred patients were randomly assigned to ipilimumab and 202 to placebo; 399 were treated with ipilimumab and 199 with placebo. Median OS was 28.7 months (95% CI, 24.5 to 32.5 months) in the ipilimumab arm versus 29.7 months (95% CI, 26.1 to 34.2 months) in the placebo arm (hazard ratio, 1.11; 95.87% CI, 0.88 to 1.39; P = .3667). Median progression-free survival was 5.6 months in the ipilimumab arm versus 3.8 with placebo arm (hazard ratio, 0.67; 95.87% CI, 0.55 to 0.81). Exploratory analyses showed a higher prostate-specific antigen response rate with ipilimumab (23%) than with placebo (8%). Diarrhea (15%) was the only grade 3 to 4 treatment-related adverse event (AE) reported in ≥ 10% of ipilimumab-treated patients. Nine (2%) deaths occurred in the ipilimumab arm due to treatment-related AEs; no deaths occurred in the placebo arm. Immune-related grade 3 to 4 AEs occurred in 31% and 2% of patients, respectively. Conclusion Ipilimumab did not improve OS in patients with metastatic castration-resistant prostate cancer. The observed increases in progression-free survival and prostate-specific antigen response rates suggest antitumor activity in a patient subset.
••
Université de Montréal1, McMaster University2, Population Health Research Institute3, Queen's University4, Cape Breton Regional Hospital5, University of Calgary6, University of Manitoba7, University of Alberta8, Dalhousie University9, McGill University Health Centre10, Cleveland Clinic11, Ottawa Hospital Research Institute12, University of Ottawa13
TL;DR: The Canadian Cardiovascular Society Guidelines Committee and key Canadian opinion leaders believed there was a need for up to date guidelines that used the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system of evidence assessment for patients who undergo noncardiac surgery.
••
Johns Hopkins University1, Leipzig University2, Humanitas University3, Korea University4, Yale University5, West Virginia University6, University of Barcelona7, St George's, University of London8, Indiana University9, National Yang-Ming University10, Cleveland Clinic11, Aarhus University12, University at Buffalo13, Imperial College London14, Primary Children's Hospital15, Erasmus University Rotterdam16, Yeshiva University17, Ghent University18, Baylor University19, Virginia Commonwealth University20, Harvard University21, Federal University of São Paulo22, University of California, San Francisco23, Beaumont Hospital24, Boston University25, University of Oklahoma26, University of Michigan27, Carlos III Health Institute28, University of Melbourne29, Saint Louis University30, Université de Montréal31, University of Pennsylvania32, McGill University33, Mayo Clinic34, Lahey Hospital & Medical Center35, Royal Adelaide Hospital36, University of Milan37, University of Toronto38, Loyola University Chicago39, Jikei University School of Medicine40
TL;DR: This 2017 Consensus Statement is to provide a state-of-the-art review of the field of catheter and surgical ablation of AF and to report the findings of a writing group, convened by these five international societies.
Abstract: During the past three decades, catheter and surgical ablation of atrial fibrillation (AF) have evolved from investigational procedures to their current role as effective treatment options for patients with AF. Surgical ablation of AF, using either standard, minimally invasive, or hybrid techniques, is available in most major hospitals throughout the world. Catheter ablation of AF is even more widely available, and is now the most commonly performed catheter ablation procedure.
In 2007, an initial Consensus Statement on Catheter and Surgical AF Ablation was developed as a joint effort of the Heart Rhythm Society (HRS), the European Heart Rhythm Association (EHRA), and the European Cardiac Arrhythmia Society (ECAS).1 The 2007 document was also developed in collaboration with the Society of Thoracic Surgeons (STS) and the American College of Cardiology (ACC). This Consensus Statement on Catheter and Surgical AF Ablation was rewritten in 2012 to reflect the many advances in AF ablation that had occurred in the interim.2 The rate of advancement in the tools, techniques, and outcomes of AF ablation continue to increase as enormous research efforts are focused on the mechanisms, outcomes, and treatment of AF. For this reason, the HRS initiated an effort to rewrite and update this Consensus Statement. Reflecting both the worldwide importance of AF, as well as the worldwide performance of AF ablation, this document is the result of a joint partnership between the HRS, EHRA, ECAS, the Asia Pacific Heart Rhythm Society (APHRS), and the Latin American Society of Cardiac Stimulation and Electrophysiology (Sociedad Latinoamericana de Estimulacion Cardiaca y Electrofisiologia [SOLAECE]). The purpose of this 2017 Consensus Statement is to provide a state-of-the-art review of the field of catheter and surgical ablation of AF and to report the findings of a writing group, convened by these five international societies. The writing group is charged with defining the indications, techniques, and outcomes of AF ablation procedures. Included within this document are recommendations pertinent to the design of clinical trials in the field of AF ablation and the reporting of outcomes, including definitions relevant to this topic.
The writing group is composed of 60 experts representing 11 organizations: HRS, EHRA, ECAS, APHRS, SOLAECE, STS, ACC, American Heart Association (AHA), Canadian Heart Rhythm Society (CHRS), Japanese Heart Rhythm Society (JHRS), and Brazilian Society of Cardiac Arrhythmias (Sociedade Brasileira de Arritmias Cardiacas [SOBRAC]). All the members of the writing group, as well as peer reviewers of the document, have provided disclosure statements for all relationships that might be perceived as real or potential conflicts of interest. All author and peer reviewer disclosure information is provided in Appendix A and Appendix B.
In writing a consensus document, it is recognized that consensus does not mean that there was complete agreement among all the writing group members. Surveys of the entire writing group were used to identify areas of consensus concerning performance of AF ablation procedures and to develop recommendations concerning the indications for catheter and surgical AF ablation. These recommendations were systematically balloted by the 60 writing group members and were approved by a minimum of 80% of these members. The recommendations were also subject to a 1-month public comment period. Each partnering and collaborating organization then officially reviewed, commented on, edited, and endorsed the final document and recommendations.
The grading system for indication of class of evidence level was adapted based on that used by the ACC and the AHA.3,4 It is important to state, however, that this document is not a guideline. The indications for catheter and surgical ablation of AF, as well as recommendations for procedure performance, are presented with a Class and Level of Evidence (LOE) to be consistent with what the reader is familiar with seeing in guideline statements. A Class I recommendation means that the benefits of the AF ablation procedure markedly exceed the risks, and that AF ablation should be performed; a Class IIa recommendation means that the benefits of an AF ablation procedure exceed the risks, and that it is reasonable to perform AF ablation; a Class IIb recommendation means that the benefit of AF ablation is greater or equal to the risks, and that AF ablation may be considered; and a Class III recommendation means that AF ablation is of no proven benefit and is not recommended.
The writing group reviewed and ranked evidence supporting current recommendations with the weight of evidence ranked as Level A if the data were derived from high-quality evidence from more than one randomized clinical trial, meta-analyses of high-quality randomized clinical trials, or one or more randomized clinical trials corroborated by high-quality registry studies. The writing group ranked available evidence as Level B-R when there was moderate-quality evidence from one or more randomized clinical trials, or meta-analyses of moderate-quality randomized clinical trials. Level B-NR was used to denote moderate-quality evidence from one or more well-designed, well-executed nonrandomized studies, observational studies, or registry studies. This designation was also used to denote moderate-quality evidence from meta-analyses of such studies. Evidence was ranked as Level C-LD when the primary source of the recommendation was randomized or nonrandomized observational or registry studies with limitations of design or execution, meta-analyses of such studies, or physiological or mechanistic studies of human subjects. Level C-EO was defined as expert opinion based on the clinical experience of the writing group.
Despite a large number of authors, the participation of several societies and professional organizations, and the attempts of the group to reflect the current knowledge in the field adequately, this document is not intended as a guideline. Rather, the group would like to refer to the current guidelines on AF management for the purpose of guiding overall AF management strategies.5,6 This consensus document is specifically focused on catheter and surgical ablation of AF, and summarizes the opinion of the writing group members based on an extensive literature review as well as their own experience. It is directed to all health care professionals who are involved in the care of patients with AF, particularly those who are caring for patients who are undergoing, or are being considered for, catheter or surgical ablation procedures for AF, and those involved in research in the field of AF ablation. This statement is not intended to recommend or promote catheter or surgical ablation of AF. Rather, the ultimate judgment regarding care of a particular patient must be made by the health care provider and the patient in light of all the circumstances presented by that patient.
The main objective of this document is to improve patient care by providing a foundation of knowledge for those involved with catheter ablation of AF. A second major objective is to provide recommendations for designing clinical trials and reporting outcomes of clinical trials of AF ablation. It is recognized that this field continues to evolve rapidly. As this document was being prepared, further clinical trials of catheter and surgical ablation of AF were under way.
••
Valve Corporation1, Beth Israel Deaconess Medical Center2, McGill University Health Centre3, Population Health Research Institute4, University of Ottawa5, University of Toronto6, University of Manitoba7, Université de Montréal8, Harvard University9, Washington University in St. Louis10, Montreal Heart Institute11, Jewish General Hospital12, McGill University13
TL;DR: A brief 4-item scale encompassing lower-extremity weakness, cognitive impairment, anemia, and hypoalbuminemia outperformed other frailty scales and is recommended for use in this setting.
••
TL;DR: A state-of-the-art survey of the Benders Decomposition algorithm, emphasizing its use in combinatorial optimization and introducing a taxonomy of algorithmic enhancements and acceleration strategies based on the main components of the algorithm.