Text mining for contexts and relationships in cancer genomics literature.

MOTIVATION: Scientific advances build on the findings of existing research. The 2001 publication of the human genome has led to the production of huge volumes of literature exploring the context-specific functions and interactions of genes. Technology is needed to perform large-scale text mining of research papers to extract the reported actions of genes in specific experimental contexts and cell states, such as cancer, thereby facilitating the design of new therapeutic strategies. RESULTS: We present a new corpus and Text Mining methodology that can accurately identify and extract the most important details of cancer genomics experiments from biomedical texts. We build a Named Entity Recognition model that accurately extracts relevant experiment details from PubMed abstract text, and a second model that identifies the relationships between them. This system outperforms earlier models and enables the analysis of gene function in diverse and dynamically evolving experimental contexts. AVAILABILITY AND IMPLEMENTATION: Code and data are available here: https://github.com/cambridgeltl/functional-genomics-ie.

Prevalence and impact of hazardous alcohol use in intensive care cohort: A multicenter, register-based study.

BACKGROUND: Reports of the prevalence and impact of hazardous alcohol use among intensive care unit (ICU) patients are contradictory. We aimed to study the prevalence of hazardous alcohol use among ICU patients and its association with ICU length of stay (LOS) and mortality. METHODS: Finnish ICUs have been using the Alcohol Use Disorder Identification Test-Consumption (AUDIT-C) to evaluate and record patients' alcohol use into the Finnish Intensive Care Consortium's Database (FICC). We retrieved data from the FICC from a 3-month period. We excluded data from centers with an AUDIT-C recording rate of less than 70% of admissions. We defined hazardous alcohol use as a score of 5 or more for women and 6 or more for men from a maximum score of 12 points. RESULTS: Two thousand forty-five patients were treated in the 10 centers with an AUDIT-C recording rate of 70% or higher. AUDIT-C was available for 1576 (77%) patients and indicated hazardous alcohol use for 334 (21%) patients who were more often younger (median age 55 [interquartile range 42-65] vs 67 [57-74] [P < .001]) and male (78.1% vs 61.3% [P < .001]) compared to other patients. We found no difference in LOS or hospital mortality between hazardous and non-hazardous alcohol users. Among the non-abstinent, risk of death within a year increased with increasing AUDIT-C scores adjusted odds ratio 1.077 (95% confidence interval, 1.006-1.152) per point. CONCLUSION: The prevalence of hazardous alcohol use in Finnish ICUs was 21%. Patients with hazardous alcohol use were more often younger and male compared with non-hazardous alcohol users.

LION LBD: a literature-based discovery system for cancer biology.

MOTIVATION: The overwhelming size and rapid growth of the biomedical literature make it impossible for scientists to read all studies related to their work, potentially leading to missed connections and wasted time and resources. Literature-based discovery (LBD) aims to alleviate these issues by identifying implicit links between disjoint parts of the literature. While LBD has been studied in depth since its introduction three decades ago, there has been limited work making use of recent advances in biomedical text processing methods in LBD. RESULTS: We present LION LBD, a literature-based discovery system that enables researchers to navigate published information and supports hypothesis generation and testing. The system is built with a particular focus on the molecular biology of cancer using state-of-the-art machine learning and natural language processing methods, including named entity recognition and grounding to domain ontologies covering a wide range of entity types and a novel approach to detecting references to the hallmarks of cancer in text. LION LBD implements a broad selection of co-occurrence based metrics for analyzing the strength of entity associations, and its design allows real-time search to discover indirect associations between entities in a database of tens of millions of publications while preserving the ability of users to explore each mention in its original context in the literature. Evaluations of the system demonstrate its ability to identify undiscovered links and rank relevant concepts highly among potential connections. AVAILABILITY AND IMPLEMENTATION: The LION LBD system is available via a web-based user interface and a programmable API, and all components of the system are made available under open licenses from the project home page http://lbd.lionproject.net. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Neural networks for link prediction in realistic biomedical graphs: a multi-dimensional evaluation of graph embedding-based approaches.

BACKGROUND: Link prediction in biomedical graphs has several important applications including predicting Drug-Target Interactions (DTI), Protein-Protein Interaction (PPI) prediction and Literature-Based Discovery (LBD). It can be done using a classifier to output the probability of link formation between nodes. Recently several works have used neural networks to create node representations which allow rich inputs to neural classifiers. Preliminary works were done on this and report promising results. However they did not use realistic settings like time-slicing, evaluate performances with comprehensive metrics or explain when or why neural network methods outperform. We investigated how inputs from four node representation algorithms affect performance of a neural link predictor on random- and time-sliced biomedical graphs of real-world sizes (∼ 6 million edges) containing information relevant to DTI, PPI and LBD. We compared the performance of the neural link predictor to those of established baselines and report performance across five metrics. RESULTS: In random- and time-sliced experiments when the neural network methods were able to learn good node representations and there was a negligible amount of disconnected nodes, those approaches outperformed the baselines. In the smallest graph (∼ 15,000 edges) and in larger graphs with approximately 14% disconnected nodes, baselines such as Common Neighbours proved a justifiable choice for link prediction. At low recall levels (∼ 0.3) the approaches were mostly equal, but at higher recall levels across all nodes and average performance at individual nodes, neural network approaches were superior. Analysis showed that neural network methods performed well on links between nodes with no previous common neighbours; potentially the most interesting links. Additionally, while neural network methods benefit from large amounts of data, they require considerable amounts of computational resources to utilise them. CONCLUSIONS: Our results indicate that when there is enough data for the neural network methods to use and there are a negligible amount of disconnected nodes, those approaches outperform the baselines. At low recall levels the approaches are mostly equal but at higher recall levels and average performance at individual nodes, neural network approaches are superior. Performance at nodes without common neighbours which indicate more unexpected and perhaps more useful links account for this.

Bio-SimVerb and Bio-SimLex: wide-coverage evaluation sets of word similarity in biomedicine.

BACKGROUND: Word representations support a variety of Natural Language Processing (NLP) tasks. The quality of these representations is typically assessed by comparing the distances in the induced vector spaces against human similarity judgements. Whereas comprehensive evaluation resources have recently been developed for the general domain, similar resources for biomedicine currently suffer from the lack of coverage, both in terms of word types included and with respect to the semantic distinctions. Notably, verbs have been excluded, although they are essential for the interpretation of biomedical language. Further, current resources do not discern between semantic similarity and semantic relatedness, although this has been proven as an important predictor of the usefulness of word representations and their performance in downstream applications. RESULTS: We present two novel comprehensive resources targeting the evaluation of word representations in biomedicine. These resources, Bio-SimVerb and Bio-SimLex, address the previously mentioned problems, and can be used for evaluations of verb and noun representations respectively. In our experiments, we have computed the Pearson's correlation between performances on intrinsic and extrinsic tasks using twelve popular state-of-the-art representation models (e.g. word2vec models). The intrinsic-extrinsic correlations using our datasets are notably higher than with previous intrinsic evaluation benchmarks such as UMNSRS and MayoSRS. In addition, when evaluating representation models for their abilities to capture verb and noun semantics individually, we show a considerable variation between performances across all models. CONCLUSION: Bio-SimVerb and Bio-SimLex enable intrinsic evaluation of word representations. This evaluation can serve as a predictor of performance on various downstream tasks in the biomedical domain. The results on Bio-SimVerb and Bio-SimLex using standard word representation models highlight the importance of developing dedicated evaluation resources for NLP in biomedicine for particular word classes (e.g. verbs). These are needed to identify the most accurate methods for learning class-specific representations. Bio-SimVerb and Bio-SimLex are publicly available.

Cancer Hallmarks Analytics Tool (CHAT): a text mining approach to organize and evaluate scientific literature on cancer.

MOTIVATION: To understand the molecular mechanisms involved in cancer development, significant efforts are being invested in cancer research. This has resulted in millions of scientific articles. An efficient and thorough review of the existing literature is crucially important to drive new research. This time-demanding task can be supported by emerging computational approaches based on text mining which offer a great opportunity to organize and retrieve the desired information efficiently from sizable databases. One way to organize existing knowledge on cancer is to utilize the widely accepted framework of the Hallmarks of Cancer. These hallmarks refer to the alterations in cell behaviour that characterize the cancer cell. RESULTS: We created an extensive Hallmarks of Cancer taxonomy and developed automatic text mining methodology and a tool (CHAT) capable of retrieving and organizing millions of cancer-related references from PubMed into the taxonomy. The efficiency and accuracy of the tool was evaluated intrinsically as well as extrinsically by case studies. The correlations identified by the tool show that it offers a great potential to organize and correctly classify cancer-related literature. Furthermore, the tool can be useful, for example, in identifying hallmarks associated with extrinsic factors, biomarkers and therapeutics targets. AVAILABILITY AND IMPLEMENTATION: CHAT can be accessed at: http://chat.lionproject.net. The corpus of hallmark-annotated PubMed abstracts and the software are available at: http://chat.lionproject.net/about. CONTACT: simon.baker@cl.cam.ac.uk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

One- and three-year outcomes in patients treated with intermittent hemodialysis for acute kidney injury: prospective observational multicenter post-hoc FINNAKI study.

BACKGROUND: Studies reporting renal and overall survival after acute kidney injury (AKI) treated exclusively with intermittent modalities of renal replacement therapy (IRRT) are rare. This study focused on outcomes of AKI patients treated with IRRT both in intensive care units (ICUs) and non-ICU dialysis units. METHODS: This prospective observational study was carried on during a 5-month period in 17 ICUs and 17 non-ICUs. ICU and non-ICU patients (total n = 138; 65 ICU, 73 non-ICU) requiring RRT for AKI and chosen to receive IRRT were included. Patient and RRT characteristics as well as outcomes at 90 days, 1 year, and 3 years were registered. RESULTS: Characteristics of ICU and non-ICU patients differed markedly. Pre-existing chronic kidney disease (CKD) and chronic heart failure were significantly more common among non-ICU patients. At 1 year, RRT dependence was significantly more common in the non-ICU group. At 3 years, there was no significant difference between the groups either in RRT dependence or mortality. CONCLUSION: Outcome of AKI patients treated with IRRT is dismal with regard to 3-year kidney function and mortality. Although pre-existing CKD emerged as a major risk factor for end-stage renal disease after AKI, the poor kidney survival was also seen in patients without prior CKD.

Investigating the cross-lingual translatability of VerbNet-style classification.

VerbNet-the most extensive online verb lexicon currently available for English-has proved useful in supporting a variety of NLP tasks. However, its exploitation in multilingual NLP has been limited by the fact that such classifications are available for few languages only. Since manual development of VerbNet is a major undertaking, researchers have recently translated VerbNet classes from English to other languages. However, no systematic investigation has been conducted into the applicability and accuracy of such a translation approach across different, typologically diverse languages. Our study is aimed at filling this gap. We develop a systematic method for translation of VerbNet classes from English to other languages which we first apply to Polish and subsequently to Croatian, Mandarin, Japanese, Italian, and Finnish. Our results on Polish demonstrate high translatability with all the classes (96% of English member verbs successfully translated into Polish) and strong inter-annotator agreement, revealing a promising degree of overlap in the resultant classifications. The results on other languages are equally promising. This demonstrates that VerbNet classes have strong cross-lingual potential and the proposed method could be applied to obtain gold standards for automatic verb classification in different languages. We make our annotation guidelines and the six language-specific verb classifications available with this paper.

Link prediction in drug-target interactions network using similarity indices.

BACKGROUND: In silico drug-target interaction (DTI) prediction plays an integral role in drug repositioning: the discovery of new uses for existing drugs. One popular method of drug repositioning is network-based DTI prediction, which uses complex network theory to predict DTIs from a drug-target network. Currently, most network-based DTI prediction is based on machine learning - methods such as Restricted Boltzmann Machines (RBM) or Support Vector Machines (SVM). These methods require additional information about the characteristics of drugs, targets and DTIs, such as chemical structure, genome sequence, binding types, causes of interactions, etc., and do not perform satisfactorily when such information is unavailable. We propose a new, alternative method for DTI prediction that makes use of only network topology information attempting to solve this problem. RESULTS: We compare our method for DTI prediction against the well-known RBM approach. We show that when applied to the MATADOR database, our approach based on node neighborhoods yield higher precision for high-ranking predictions than RBM when no information regarding DTI types is available. CONCLUSION: This demonstrates that approaches purely based on network topology provide a more suitable approach to DTI prediction in the many real-life situations where little or no prior knowledge is available about the characteristics of drugs, targets, or their interactions.

A neural network multi-task learning approach to biomedical named entity recognition.

BACKGROUND: Named Entity Recognition (NER) is a key task in biomedical text mining. Accurate NER systems require task-specific, manually-annotated datasets, which are expensive to develop and thus limited in size. Since such datasets contain related but different information, an interesting question is whether it might be possible to use them together to improve NER performance. To investigate this, we develop supervised, multi-task, convolutional neural network models and apply them to a large number of varied existing biomedical named entity datasets. Additionally, we investigated the effect of dataset size on performance in both single- and multi-task settings. RESULTS: We present a single-task model for NER, a Multi-output multi-task model and a Dependent multi-task model. We apply the three models to 15 biomedical datasets containing multiple named entities including Anatomy, Chemical, Disease, Gene/Protein and Species. Each dataset represent a task. The results from the single-task model and the multi-task models are then compared for evidence of benefits from Multi-task Learning. With the Multi-output multi-task model we observed an average F-score improvement of 0.8% when compared to the single-task model from an average baseline of 78.4%. Although there was a significant drop in performance on one dataset, performance improves significantly for five datasets by up to 6.3%. For the Dependent multi-task model we observed an average improvement of 0.4% when compared to the single-task model. There were no significant drops in performance on any dataset, and performance improves significantly for six datasets by up to 1.1%. The dataset size experiments found that as dataset size decreased, the multi-output model's performance increased compared to the single-task model's. Using 50, 25 and 10% of the training data resulted in an average drop of approximately 3.4, 8 and 16.7% respectively for the single-task model but approximately 0.2, 3.0 and 9.8% for the multi-task model. CONCLUSIONS: Our results show that, on average, the multi-task models produced better NER results than the single-task models trained on a single NER dataset. We also found that Multi-task Learning is beneficial for small datasets. Across the various settings the improvements are significant, demonstrating the benefit of Multi-task Learning for this task.

Automatic semantic classification of scientific literature according to the hallmarks of cancer.

MOTIVATION: The hallmarks of cancer have become highly influential in cancer research. They reduce the complexity of cancer into 10 principles (e.g. resisting cell death and sustaining proliferative signaling) that explain the biological capabilities acquired during the development of human tumors. Since new research depends crucially on existing knowledge, technology for semantic classification of scientific literature according to the hallmarks of cancer could greatly support literature review, knowledge discovery and applications in cancer research. RESULTS: We present the first step toward the development of such technology. We introduce a corpus of 1499 PubMed abstracts annotated according to the scientific evidence they provide for the 10 currently known hallmarks of cancer. We use this corpus to train a system that classifies PubMed literature according to the hallmarks. The system uses supervised machine learning and rich features largely based on biomedical text mining. We report good performance in both intrinsic and extrinsic evaluations, demonstrating both the accuracy of the methodology and its potential in supporting practical cancer research. We discuss how this approach could be developed and applied further in the future. AVAILABILITY AND IMPLEMENTATION: The corpus of hallmark-annotated PubMed abstracts and the software for classification are available at: http://www.cl.cam.ac.uk/∼sb895/HoC.html. CONTACT: simon.baker@cl.cam.ac.uk.

Acute Kidney Injury After Cardiac Surgery by Complete KDIGO Criteria Predicts Increased Mortality.

OBJECTIVES: Acute kidney injury (AKI) occurs frequently after cardiac surgery and is associated with increased mortality. The Kidney Disease: Improving Global Outcomes (KDIGO) criteria for diagnosing AKI include creatinine and urine output values. However, the value of the latter is debated. The authors aimed to evaluate the incidence of AKI after cardiac surgery and the independent association of KDIGO criteria, especially the urine output criterion, and 2.5-year mortality. DESIGN: Prospective, observational, cohort study. SETTING: Single-center study in a university hospital. PARTICIPANTS: The study comprised 638 cardiac surgical patients from September 1, 2011, to June 20, 2012. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Hourly urine output, daily plasma creatinine, risk factors for AKI, and variables for EuroSCORE II were recorded. AKI occurred in 183 (28.7%) patients. Patients with AKI diagnosed using only urine output had higher 2.5-year mortality than did patients without AKI (9/53 [17.0%] v 23/455 [5.1%], p = 0.001). AKI was associated with mortality (hazard ratios [95% confidence intervals]: 3.3 [1.8-6.1] for KDIGO 1; 5.8 [2.7-12.1] for KDIGO 2; and 7.9 [3.5-17.6]) for KDIGO 3. KDIGO stages and AKI diagnosed using urine output were associated with mortality even after adjusting for mortality risk assessed using EuroSCORE II and risk factors for AKI. CONCLUSIONS: AKI diagnosed using only the urine output criterion without fulfilling the creatinine criterion and all stages of AKI were associated with long-term mortality. Preoperatively assessed mortality risk using EuroSCORE II did not predict this AKI-associated mortality.

Gender differences in cancer susceptibility: role of oxidative stress.

Cancer is a leading cause of death worldwide and environmental factors, including chemicals, have been suggested as major etiological incitements. Cancer statistics indicates that men get more cancer than women. However, differences in the known risk factors including life style or occupational exposure only offer partial explanation. Using a text mining tool, we have investigated the scientific literature concerning male- and female-specific rat carcinogens that induced tumors only in one gender in NTP 2-year cancer bioassay. Our evaluation shows that oxidative stress, although frequently reported for both male- and female-specific rat carcinogens, was mentioned significantly more in literature concerning male-specific rat carcinogens. Literature analysis of testosterone and estradiol showed the same pattern. Tox21 high-throughput assay results, although showing only weak association of oxidative stress-related processes for male- and female-specific rat carcinogens, provide additional support. We also analyzed the literature concerning 26 established human carcinogens (IARC group 1). Oxidative stress was more frequently reported for the majority of these carcinogens, and the Tox21 data resembled that of male-specific rat carcinogens. Thus, our data, based on about 600000 scientific abstracts and Tox21 screening assays, suggest a link between male-specific carcinogens, testosterone and oxidative stress. This implies that a different cellular response to oxidative stress in men and women may be a critical factor in explaining the greater cancer susceptibility observed in men. Although the IARC carcinogens are classified as human carcinogens, their classification largely based on epidemiological evidence from male cohorts, which raises the question whether carcinogen classifications should be gender specific.

Unsupervised discovery of information structure in biomedical documents.

MOTIVATION: Information structure (IS) analysis is a text mining technique, which classifies text in biomedical articles into categories that capture different types of information, such as objectives, methods, results and conclusions of research. It is a highly useful technique that can support a range of Biomedical Text Mining tasks and can help readers of biomedical literature find information of interest faster, accelerating the highly time-consuming process of literature review. Several approaches to IS analysis have been presented in the past, with promising results in real-world biomedical tasks. However, all existing approaches, even weakly supervised ones, require several hundreds of hand-annotated training sentences specific to the domain in question. Because biomedicine is subject to considerable domain variation, such annotations are expensive to obtain. This makes the application of IS analysis across biomedical domains difficult. In this article, we investigate an unsupervised approach to IS analysis and evaluate the performance of several unsupervised methods on a large corpus of biomedical abstracts collected from PubMed. RESULTS: Our best unsupervised algorithm (multilevel-weighted graph clustering algorithm) performs very well on the task, obtaining over 0.70 F scores for most IS categories when applied to well-known IS schemes. This level of performance is close to that of lightly supervised IS methods and has proven sufficient to aid a range of practical tasks. Thus, using an unsupervised approach, IS could be applied to support a wide range of tasks across sub-domains of biomedicine. We also demonstrate that unsupervised learning brings novel insights into IS of biomedical literature and discovers information categories that are not present in any of the existing IS schemes. AVAILABILITY AND IMPLEMENTATION: The annotated corpus and software are available at http://www.cl.cam.ac.uk/∼dk427/bio14info.html.

Frost hardiness of mycorrhizal and non-mycorrhizal Scots pine under two fertilization treatments.

Survival and functioning of mycorrhizal associations at low temperatures are not known well. In an earlier study, ectomycorrhizas did not affect the frost hardiness of Scots pine (Pinus sylvestris L.) roots, but here we studied whether differential nutrient availability would change the result and additionally, alter frost hardiness aboveground. The aim in this experiment was to compare the frost hardiness of roots and needles of mycorrhizal (Hebeloma sp.) and non-mycorrhizal Scots pine seedlings raised using two fertilization treatments and two cold-hardening regimes. The fertilization treatments were low (LF) and high (HF) application of a complete nutrient solution. Three hundred mycorrhizal and non-mycorrhizal seedlings were cultivated in growth chambers in four blocks for 16 weeks. For the first 9 weeks, the seedlings grew in long-day and high-temperature (LDHT) with low fertilization and then they were raised for 3 weeks in LDHT with either low or high fertilization. After this, half of the plants in each treatment combination remained in LDHT, and half were transferred to short-day and low-temperature (SDLT) conditions to cold acclimatize. The frost hardiness of the roots and needles was assessed using controlled freezing tests followed by electrolyte leakage tests (REL). Mycorrhizal roots were slightly more frost hardy than non-mycorrhizal roots, but only in the growing-season conditions (LDHT) in low-nutrient treatment. In LDHT and LF, the frost hardiness of the non-mycorrhizal roots was about -9 °C, and that of the non-mycorrhizal HF roots and the mycorrhizal roots in both fertilization levels was about -11 °C. However, no difference was found in the roots within the SDLT regime, and in needles, there was no difference between mycorrhizal and fertilization treatments. The frost hardiness of needles increased by SDLT treatment, being -8.5 and -14.1 °C in LDHT and SDLT, respectively. The dry mass of roots, stems, and needles was lower in LF than in HF and lower in SDLT than in LDHT. Mycorrhizal treatment did not affect the dry mass or its allocation. Although the mycorrhizal roots were slightly more frost hardy in the growing-season conditions, this is not likely to have significance in the field.

The attributable mortality of acute kidney injury: a sequentially matched analysis*.

OBJECTIVE: Acute kidney injury in the critically ill is an independent risk factor for adverse outcome. The magnitude of the impact of acute kidney injury on outcome, however, is still unclear. This study aimed to estimate the excess mortality attributable to acute kidney injury. DESIGN: We performed a sequentially matched analysis according to the day of acute kidney injury diagnosis after ICU admission. Patients with acute kidney injury and those without acute kidney injury were matched according to age, sex, ICU admission diagnosis, Simplified Acute Physiology Score II without renal and age components, and the propensity to develop acute kidney injury at each of the four matching time points. SETTING: Cohort of 16 participating ICUs from the prospective Finnish Acute Kidney Injury study. PATIENTS: Cohort of 2,719 consecutive patients with either emergency admission or elective postsurgical patients with an expected ICU stay greater than 24 hours. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Of the 2,719 patients included in the study, acute kidney injury developed in 1,081 patients (39.8%) according to the Kidney Disease: Improving Global Outcomes-definition during ICU treatment on days 1-5. Of these, 477 patients were successfully matched to 477 patients who did not develop acute kidney injury. The 90-day mortality of the matched patients with acute kidney injury was 125 of 477 (26.2%) compared with 84 of 477 (17.6%) for their matched controls without acute kidney injury. Thus, the absolute excess 90-day mortality attributable to acute kidney injury was estimated at 8.6 percentage points (95% CI, 2.6-17.6 percentage points). The population attributable risk (95% CI) of 90-day mortality associated with acute kidney injury was 19.6% (10.3-34.1%). CONCLUSIONS: In general ICU patients, the absolute excess 90-day mortality statistically attributable to acute kidney injury is substantial (8.6%), and the population attributable risk was nearly 20%. Our findings are useful in planning suitably powered future clinical trials to prevent and treat acute kidney injury in critically ill patients.

Active learning-based information structure analysis of full scientific articles and two applications for biomedical literature review.

MOTIVATION: Techniques that are capable of automatically analyzing the information structure of scientific articles could be highly useful for improving information access to biomedical literature. However, most existing approaches rely on supervised machine learning (ML) and substantial labeled data that are expensive to develop and apply to different sub-fields of biomedicine. Recent research shows that minimal supervision is sufficient for fairly accurate information structure analysis of biomedical abstracts. However, is it realistic for full articles given their high linguistic and informational complexity? We introduce and release a novel corpus of 50 biomedical articles annotated according to the Argumentative Zoning (AZ) scheme, and investigate active learning with one of the most widely used ML models-Support Vector Machines (SVM)-on this corpus. Additionally, we introduce two novel applications that use AZ to support real-life literature review in biomedicine via question answering and summarization. RESULTS: We show that active learning with SVM trained on 500 labeled sentences (6% of the corpus) performs surprisingly well with the accuracy of 82%, just 2% lower than fully supervised learning. In our question answering task, biomedical researchers find relevant information significantly faster from AZ-annotated than unannotated articles. In the summarization task, sentences extracted from particular zones are significantly more similar to gold standard summaries than those extracted from particular sections of full articles. These results demonstrate that active learning of full articles' information structure is indeed realistic and the accuracy is high enough to support real-life literature review in biomedicine. AVAILABILITY: The annotated corpus, our AZ classifier and the two novel applications are available at http://www.cl.cam.ac.uk/yg244/12bioinfo.html

Variation in the use of renal replacement therapy in patients with septic shock: a substudy of the prospective multicenter observational FINNAKI study.

INTRODUCTION: Indications for renal replacement therapy (RRT) have not been generally standardized and vary among intensive care units (ICUs). We aimed to assess the proportion, indications, and modality of RRT, as well as the association between the proportion of RRT use and 90-day mortality in patients with septic shock in Finnish adult ICUs. METHODS: We identified patients with septic shock from the prospective observational multicenter FINNAKI study conducted between 1 September 2011 and 1 February 2012. We divided the ICUs into high-RRT and low-RRT ICUs according to the median of the proportion of RRT-treated patients with septic shock. Differences in indications, and modality of RRT between ICU groups were assessed. Finally, we performed an adjusted logistic regression analysis to evaluate the possible association of the ICU group (high vs. low-RRT) with 90-day mortality. RESULTS: Of the 726 patients with septic shock, 131 (18.0%, 95% CI 15.2 to 20.9%) were treated with RRT. The proportion of RRT-treated patients varied from 3% up to 36% (median 19%) among ICUs. High-RRT ICUs included nine ICUs (354 patients) and low-RRT ICUs eight ICUs (372 patients). In the high-RRT ICUs patients with septic shock were older (P = 0.04), had more cardiovascular (P <0.001) and renal failures (P = 0.003) on the first day in the ICU, were more often mechanically ventilated, and received higher maximum doses of norepinephrine (0.25 µg/kg/min vs. 0.18 µg/kg/min, P <0.001) than in the low-RRT ICUs. No significant differences in indications for or modality of RRT existed between the ICU groups. The crude 90-day mortality rate for patients with septic shock was 36.2% (95% CI 31.1 to 41.3%) in the high-RRT ICUs compared to 33.9% (95% CI 29.0 to 38.8%) in the low-RRT ICUs, P = 0.5. In an adjusted logistic regression analysis the ICU group (high-RRT or low-RRT ICUs) was not associated with 90-day mortality. CONCLUSIONS: Patients with septic shock in ICUs with a high proportion of RRT had more severe organ dysfunctions and received more organ-supportive treatments. Importantly, the ICU group (high-RRT or low-RRT group) was not associated with 90-day mortality.

The urine protein NGAL predicts renal replacement therapy, but not acute kidney injury or 90-day mortality in critically ill adult patients.

BACKGROUND: Urine neutrophil gelatinase-associated lipocalin (uNGAL) is increasingly used as a biomarker for acute kidney injury (AKI). However, the clinical value of uNGAL with respect to AKI, renal replacement therapy (RRT), or 90-day mortality in critically ill patients is unclear. Accordingly, we tested the hypothesis that uNGAL is a clinically relevant biomarker for these end points in a large, nonselected cohort of critically ill adult patients. METHODS: We prospectively obtained urine samples from 1042 adult patients admitted to 15 Finnish intensive care units. We analyzed 3 samples (on admission, at 12 hours, and at 24 hours) with NGAL ELISA Rapid Kits (BioPorto® Diagnostics, Gentofte, Denmark). We chose the highest uNGAL (uNGAL24) for statistical analyses. We calculated the areas under receiver operating characteristics curves (AUC) with 95% confidence intervals (95% CIs), the best cutoff points with the Youden index, positive likelihood ratios (LR+), continuous net reclassification improvement (NRI), and the integrated discrimination improvement (IDI). We performed sensitivity analyses excluding patients with AKI or RRT on day 1, sepsis, or with missing baseline serum creatinine concentration. RESULTS: In this study population, the AUC of uNGAL24 (95% CI) for development of AKI (defined by the Kidney Disease: Improving Global Outcomes [KDIGO] criteria) was 0.733 (0.701-0.765), and the continuous NRI for AKI was 56.9%. For RRT, the AUC of uNGAL24 (95% CI) was 0.839 (0.797-0.880), and NRI 56.3%. For 90-day mortality, the AUC of uNGAL24 (95% CI) was 0.634 (0.593 to 0.675), and NRI 15.3%. The LR+ (95% CI) for RRT was 3.81 (3.26-4.47). CONCLUSION: In this study, we found that uNGAL associated well with the initiation of RRT but did not provide additional predictive value regarding AKI or 90-day mortality in critically ill patients.

Protein inhibitor of activated STAT3 (PIAS3) protein promotes SUMOylation and nuclear sequestration of the intracellular domain of ErbB4 protein.

ErbB4 is a receptor tyrosine kinase implicated in the development and homeostasis of the heart, central nervous system, and mammary gland. Cleavable isoforms of ErbB4 release a soluble intracellular domain (ICD) that can translocate to the nucleus and function as a transcriptional coregulator. In search of regulatory mechanisms of ErbB4 ICD function, we identified PIAS3 as a novel interaction partner of ErbB4 ICD. In keeping with the small ubiquitin-like modifier (SUMO) E3 ligase function of protein inhibitor of activated STAT (PIAS) proteins, we showed that the ErbB4 ICD is modified by SUMO, and that PIAS3 stimulates the SUMOylation. Upon overexpression of PIAS3, the ErbB4 ICD generated from the full-length receptor accumulated into the nucleus in a manner that was dependent on the functional nuclear localization signal of ErbB4. In the nucleus, ErbB4 colocalized with PIAS3 and SUMO-1 in promyelocytic leukemia nuclear bodies, nuclear domains involved in regulation of transcription. Accordingly, PIAS3 overexpression had an effect on the transcriptional coregulatory activity of ErbB4, repressing its ability to coactivate transcription with Yes-associated protein. Finally, knockdown of PIAS3 with siRNA partially rescued the inhibitory effect of the ErbB4 ICD on differentiation of MDA-MB-468 breast cancer and HC11 mammary epithelial cells. Our findings illustrate that PIAS3 is a novel regulator of ErbB4 receptor tyrosine kinase, controlling its nuclear sequestration and function.