Search

Cellular therapy to treat heart failure is an ongoing focus of intense research, but progress toward structural and functional recovery remains modest. Engineered augmentation of established cellular effectors overcomes impediments to enhance reparative activity. Such ‘next generation’ implementation includes delivery of combinatorial cell populations exerting synergistic effects. Concurrent isolation and expansion of three distinct cardiac-derived interstitial cell types from human heart tissue, previously reported by our group, prompted design of a 3D structure that maximizes cellular interaction, allows for defined cell ratios, controls size, enables injectability, and minimizes cell loss. Herein, mesenchymal stem cells (MSCs), endothelial progenitor cells (EPCs) and c-Kit+ cardiac interstitial cells (cCICs) when cultured together spontaneously form scaffold-free 3D microenvironments termed CardioClusters. scRNA-Seq profiling reveals CardioCluster expression of stem cell-relevant factors, adhesion/extracellular-matrix molecules, and cytokines, while maintaining a more native transcriptome similar to endogenous cardiac cells. CardioCluster intramyocardial delivery improves cell retention and capillary density with preservation of cardiomyocyte size and long-term cardiac function in a murine infarction model followed 20 weeks. CardioCluster utilization in this preclinical setting establish fundamental insights, laying the framework for optimization in cell-based therapeutics intended to mitigate cardiomyopathic damage., Despite recent progress to advance cardiac cell-based therapy for patients, heart failure mortality rivals most cancers. Here, the authors describe an approach to control and pattern 3 distinct human cardiac cell populations to promote superior repair and regeneration after myocardial infarction.

Data augmentation is a ubiquitous technique for increasing the size of labeled training sets by leveraging task-specific data transformations that preserve class labels. While it is often easy for domain experts to specify individual transformations, constructing and tuning the more sophisticated compositions typically needed to achieve state-of-the-art results is a time-consuming manual task in practice. We propose a method for automating this process by learning a generative sequence model over user-specified transformation functions using a generative adversarial approach. Our method can make use of arbitrary, non-deterministic transformation functions, is robust to misspecified user input, and is trained on unlabeled data. The learned transformation model can then be used to perform data augmentation for any end discriminative model. In our experiments, we show the efficacy of our approach on both image and text datasets, achieving improvements of 4.0 accuracy points on CIFAR-10, 1.4 F1 points on the ACE relation extraction task, and 3.4 accuracy points when using domain-specific transformation operations on a medical imaging dataset as compared to standard heuristic augmentation approaches.

State-of-the art machine learning methods such as deep learning rely on large sets of hand-labeled training data. Collecting training data is prohibitively slow and expensive, especially when technical domain expertise is required; even the largest technology companies struggle with this challenge. We address this critical bottleneck with Snorkel, a new system for quickly creating, managing, and modeling training sets. Snorkel enables users to generate large volumes of training data by writing labeling functions, which are simple functions that express heuristics and other weak supervision strategies. These user-authored labeling functions may have low accuracies and may overlap and conflict, but Snorkel automatically learns their accuracies and synthesizes their output labels. Experiments and theory show that surprisingly, by modeling the labeling process in this way, we can train high-accuracy machine learning models even using potentially lower-accuracy inputs. Snorkel is currently used in production at top technology and consulting companies, and used by researchers to extract information from electronic health records, after-action combat reports, and the scientific literature. In this demonstration, we focus on the challenging task of information extraction, a common application of Snorkel in practice. Using the task of extracting corporate employment relationships from news articles, we will demonstrate and build intuition for a radically different way of developing machine learning systems which allows us to effectively bypass the bottleneck of hand-labeling training data.

Labeling training data is increasingly the largest bottleneck in deploying machine learning systems. We present Snorkel, a first-of-its-kind system that enables users to train state-of-the-art models without hand labeling any training data. Instead, users write labeling functions that express arbitrary heuristics, which can have unknown accuracies and correlations. Snorkel denoises their outputs without access to ground truth by incorporating the first end-to-end implementation of our recently proposed machine learning paradigm, data programming. We present a flexible interface layer for writing labeling functions based on our experience over the past year collaborating with companies, agencies, and research laboratories. In a user study, subject matter experts build models \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.8\times $$\end{document}2.8× faster and increase predictive performance an average \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$45.5\%$$\end{document}45.5% versus seven hours of hand labeling. We study the modeling trade-offs in this new setting and propose an optimizer for automating trade-off decisions that gives up to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$1.8\times $$\end{document}1.8× speedup per pipeline execution. In two collaborations, with the US Department of Veterans Affairs and the US Food and Drug Administration, and on four open-source text and image data sets representative of other deployments, Snorkel provides \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$132\%$$\end{document}132% average improvements to predictive performance over prior heuristic approaches and comes within an average \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$3.60\%$$\end{document}3.60% of the predictive performance of large hand-curated training sets.

Populating large-scale structured databases from unstructured sources is a critical and challenging task in data analytics. As automated feature engineering methods grow increasingly prevalent, constructing sufficiently large labeled training sets has become the primary hurdle in building machine learning information extraction systems. In light of this, we have taken a new approach called data programming [7]. Rather than hand-labeling data, in the data programming paradigm, users generate large amounts of noisy training labels by programmatically encoding domain heuristics as simple rules. Using this approach over more traditional distant supervision methods and fully supervised approaches using labeled data, we have been able to construct knowledge base systems more rapidly and with higher quality. Since the ability to quickly prototype, evaluate, and debug these rules is a key component of this paradigm, we introduce DDLite, an interactive development framework for data programming. This paper reports feedback collected from DDLite users across a diverse set of entity extraction tasks. We share observations from several DDLite hackathons in which 10 biomedical researchers prototyped information extraction pipelines for chemicals, diseases, and anatomical named entities. Initial results were promising, with the disease tagging team obtaining an F1 score within 10 points of the state-of-the-art in only a single day-long hackathon's work. Our key insights concern the challenges of writing diverse rule sets for generating labels, and exploring training data. These findings motivate several areas of active data programming research.

Objectives: Advances in multiparametric magnetic resonance imaging (mpMRI) and ultrasound/MRI fusion imaging offer a powerful alternative to the typical undirected approach to diagnosing prostate cancer. However, these methods require the time and expertise needed to interpret mpMRI image scenes. In this paper, a machine learning framework for automatically detecting and localizing cancerous lesions within the prostate is developed and evaluated. Methods: Two studies were performed to gather MRI and pathology data. The 12 patients in the first study underwent an MRI session to obtain structural, diffusion-weighted, and dynamic contrast enhanced image vol- umes of the prostate, and regions suspected of being cancerous from the MRI data were manually contoured by radiologists. Whole-mount slices of the prostate were obtained for the patients in the second study, in addition to structural and diffusion-weighted MRI data, for pathology verification. A 3-D feature set for voxel-wise appear- ance description combining intensity data, textural operators, and zonal approximations was generated. Voxels in a test set were classified as normal or cancer using a decision forest-based model initialized using Gaussian discriminant analysis. A leave-one-patient-out cross-validation scheme was used to assess the predictions against the expert manual segmentations confirmed as cancer by biopsy. Results: We achieved an area under the average receiver-operator characteristic curve of 0.923 for the first study, and visual assessment of the probability maps showed 21 out of 22 tumors were identified while a high level of specificity was maintained. In addition to evaluating the model against related approaches, the effects of the individual MRI parameter types were explored, and pathological verification using whole-mount slices from the second study was performed. Conclusions: The results of this paper show that the combination of mpMRI and machine learning is a powerful tool for quantitatively diagnosing prostate cancer.

Zur gefaschirurgischen Versorgung eines Verschlusses der A. brachialis wurde bei einem 72-jahrigen Mann eine vertikale infraklavikulare Plexusanasthesie (VIP) durchgefuhrt. 20 Stunden postoperativ fiel eine komplette Parese der betroffenen Extremitat auf. Ein erneuter Verschluss konnte ausgeschlossen werden. Wahrend der Umlagerung im CT bemerkte der Patient ein knackendes Gerausch im HWS-Bereich mit schlagartig einsetzender Besserung der neurologischen Symptome. Der radiologische Befund ergab keinen Hinweis auf eine zerebrale Ischamie oder Plexuslasion. Als Zufallsbefund zeigte sich ein ausgepragter zervikaler Bandscheibenvorfall mit Kompression des Myelons. Erfreulicherweise kam es bis zum zweiten postoperativen Tag bezuglich der neurologischen Defizite zur einer Restitutio ad integrum.

Background Metastatic colorectal cancer (mCRC) is commonly treated with 5-fluorouracil, folinic acid, and oxaliplatin or irinotecan. The multitargeted kinase inhibitor, regorafenib, was combined with chemotherapy as first- or second-line treatment of mCRC to assess safety and pharmacokinetics (primary objectives) and tumor response (secondary objective). Patients and methods Forty-five patients were treated every 2 weeks with 5-fluorouracil 400 mg/m2 bolus then 2400 mg/m2 over 46 h, folinic acid 400 mg/m2, and either oxaliplatin 85 mg/m2 or irinotecan 180 mg/m2. On days 4–10, patients received regorafenib 160 mg orally once daily. Results The median duration of treatment was 108 (range 2–345 days). Treatment was stopped for adverse events or death (17 patients), disease progression (11 patients), and consent withdrawal or investigator decision (11 patients). Six patients remained on regorafenib at data cutoff (two without chemotherapy). Drug-related adverse events occurred in 44 patients [grade ≥3 in 32 patients: mostly neutropenia (17 patients) and leukopenia, hand–foot skin reaction, and hypophosphatemia (four patients each)]. Thirty-three patients achieved disease control (partial response or stable disease) for a median of 126 (range 42–281 days). Conclusion Regorafenib had acceptable tolerability in combination with chemotherapy, with increased exposure of irinotecan and SN-38 but no significant effect on 5-fluorouracil or oxaliplatin pharmacokinetics.

A 72-year-old man with an obliteration of the brachial artery received a vertical infraclavicular block (VIP) for vascular surgery but 20 h after the operation a complete paresis of the affected extremity occurred. A new vascular obliteration could be excluded. During the diagnostic examination the patient noticed a snapping noise in the cervical column when moving his head and an abrupt recovery of the neurological deficits occurred. The radiological diagnostic provided no indication of cerebral ischemia or lesions of the brachial plexus. An additional diagnostic finding was a profound herniated vertebral disc with compression of the myelon. Fortunately, the neurological deficits completely returned to normal.

3585 Background: Regorafenib is an oral multikinase inhibitor of angiogenic (VEGFR1-3, TIE2), stromal (PDGFR-β, FGFR), and oncogenic kinases (KIT, RET, B-RAF). Regorafenib resulted in a 74% disease...

It is weU known that the March, 1902 publication of Lenin's What is to Be Done! laid many of the ideological foundations of Bolshevism, foundations which began to assume a definite organizational form at the Second Congress of the Russian Social-Democratic Labor Party in the Summer of the foUowing year. The distinction between consciousness and spontaneity, the conception of a centralized vanguard party, the emphasis on consistent political activity based on agitation and propaganda, the 'plan for an aU-Russian political newspaper' aU these insistently repeated themes of Lenin's most influential work would provide much of the material around which Russian revolutiona ries would organize, argue, and divide in the coming years.1 Although What is to Be Donel expressed these motifs in a particularly sharp and polemical fashion, the foundation on which Lenin's pamphlet was based was largely in place before 1902.2 It was centered around his insistence that the 'party of a new type', whose construction he considered to be the 'chief task' facing Russian Social-Democracy, be organized around the regular printing, distribution, and use of a poUtical paper. I propose to examine the ideological and poUtical foundations of Lenin's theory of the party press, taking as my starting-point his often-repeated assertion that the press was the center around which his thinking revolved prior to the pubUcation of What is to Be Donel Several analysts have devoted considerable attention to Lenin's thoughts concerning the press and have tried to connect them to his concern with party-bu?ding.3 Others consider the role of the press in a relatively brief, superficial and unrevealing manner.4 But almost aU existing accounts tend to treat the press either as a strict organizational question related to specific tactics in the struggle against tsarism5 or as a function of Lenin's personal, poUtical drive for leadership, domination and power.6 The result is that the ideological and poUtical roots of his concerns tend to be obscured or ignored, and this examination represents an attempt to clarify both the reasoning behind his projections and the role he assigned to the paper. Wh?e his thinking was partly taUored to the specific problems facing Russian Social-Democracy

The excavation of two adjacent round barrows at Trelystan, Long Mountain, Powys, in 1979, has revealed a complex sequence of Late Neolithic and Early Bronze Age funerary structures and traces of Late Neolithic settlement. The earliest structure was a large pit grave, dated to about 2400 bc, which was superseded at about 2200 bc by a settlement represented by stake-walled buildings associated with Grooved Ware. Following this some activity took place, possibly domestic, which is represented by sherds probably derived from several southern Beakers, which by analogy with sites elsewhere are to be dated to a period after about 1850 bc. The subsequent Bronze Age cemetery, dated to between about 1800 and 1500 bc (but possibly continuing later), presents a sequence of burial types and structures which can broadly be seen to illustrate a change from the concept of barrow cemetery to that of cemetery barrow. The earliest burials, which consist of cremations in pits and occasionally accompanied by a Food Vessel, were covered by separate small mounds of stone or turf, or a combination of the two. These were eventually overlain by and amalgamated beneath two larger turf barrows associated with Food Vessel Urns, which employed stake circles in their construction and which acted as repositories for additional cremation burials. The cemetery was sited along a pre-existing boundary fence, set up after the Late Neolithic phase, which was renewed at various times throughout the life of the cemetery in response to changes in its layout. A small undated cemetery of inhumation graves, possibly of the Early Christian period, was founded on the eastern side of the barrows at a later date.

SummaryAbout three dozen bronze anvils are known from Bronze Age Europe. They are small in size, probably indicating that they were used for delicate work. Simple, Beaked and Complex forms are distinguished, based on the presence of special features such as work surfaces, beaks, swages and punching holes, which were used in various processes of manufacture and decoration of small bronze and gold artefacts. The anvils are widely distributed throughout north-west Europe, and dated examples belong to the later Middle and Late Bronze Age (c. 1200–700 B.C.). Thirty-six anvils are catalogued and briefly described.