Search

Cohort selection is an essential prerequisite for clinical research, determining whether an individual satisfies given selection criteria. Previous works for cohort selection usually treated each selection criterion independently and ignored not only the meaning of each selection criterion but the relations among cohort selection criteria. To solve the problems above, we propose a novel unified machine reading comprehension (MRC) framework. In this MRC framework, we design simple rules to generate questions for each criterion from cohort selection guidelines and treat clues extracted by trigger words from patients' medical records as passages. A series of state-of-the-art MRC models based on BiDAF, BIMPM, BERT, BioBERT, NCBI-BERT, and RoBERTa are deployed to determine which question and passage pairs match. We also introduce a cross-criterion attention mechanism on representations of question and passage pairs to model relations among cohort selection criteria. Results on two datasets, that is, the dataset of the 2018 National NLP Clinical Challenge (N2C2) for cohort selection and a dataset from the MIMIC-III dataset, show that our NCBI-BERT MRC model with cross-criterion attention mechanism achieves the highest micro-averaged F1-score of 0.9070 on the N2C2 dataset and 0.8353 on the MIMIC-III dataset. It is competitive to the best system that relies on a large number of rules defined by medical experts on the N2C2 dataset. Comparing these two models, we find that the NCBI-BERT MRC model mainly performs worse on mathematical logic criteria. When using rules instead of the NCBI-BERT MRC model on some criteria regarding mathematical logic on the N2C2 dataset, we obtain a new benchmark with an F1-score of 0.9163, indicating that it is easy to integrate rules into MRC models for improvement.

Background Drug repurposing is to find new indications of approved drugs, which is essential for investigating new uses for approved or investigational drug efficiency. The active gene annotation corpus (named AGAC) is annotated by human experts, which was developed to support knowledge discovery for drug repurposing. The AGAC track of the BioNLP Open Shared Tasks using this corpus is organized by EMNLP-BioNLP 2019, where the “Selective annotation” attribution makes AGAC track more challenging than other traditional sequence labeling tasks. In this work, we show our methods for trigger word detection (Task 1) and its thematic role identification (Task 2) in the AGAC track. As a step forward to drug repurposing research, our work can also be applied to large-scale automatic extraction of medical text knowledge. Methods To meet the challenges of the two tasks, we consider Task 1 as the medical name entity recognition (NER), which cultivates molecular phenomena related to gene mutation. And we regard Task 2 as a relation extraction task, which captures the thematic roles between entities. In this work, we exploit pre-trained biomedical language representation models (e.g., BioBERT) in the information extraction pipeline for mutation-disease knowledge collection from PubMed. Moreover, we design the fine-tuning framework by using a multi-task learning technique and extra features. We further investigate different approaches to consolidate and transfer the knowledge from varying sources and illustrate the performance of our model on the AGAC corpus. Our approach is based on fine-tuned BERT, BioBERT, NCBI BERT, and ClinicalBERT using multi-task learning. Further experiments show the effectiveness of knowledge transformation and the ensemble integration of models of two tasks. We conduct a performance comparison of various algorithms. We also do an ablation study on the development set of Task 1 to examine the effectiveness of each component of our method. Results Compared with competitor methods, our model obtained the highest Precision (0.63), Recall (0.56), and F-score value (0.60) in Task 1, which ranks first place. It outperformed the baseline method provided by the organizers by 0.10 in F-score. The model shared the same encoding layers for the named entity recognition and relation extraction parts. And we obtained a second high F-score (0.25) in Task 2 with a simple but effective framework. Conclusions Experimental results on the benchmark annotation of genes with active mutation-centric function changes corpus show that integrating pre-trained biomedical language representation models (i.e., BERT, NCBI BERT, ClinicalBERT, BioBERT) into a pipe of information extraction methods with multi-task learning can improve the ability to collect mutation-disease knowledge from PubMed.

Current models for event causality identification (ECI) mainly adopt a supervised framework, which heavily rely on labeled data for training. Unfortunately, the scale of current annotated datasets is relatively limited, which cannot provide sufficient support for models to capture useful indicators from causal statements, especially for handing those new, unseen cases. To alleviate this problem, we propose a novel approach, shortly named CauSeRL, which leverages external causal statements for event causality identification. First of all, we design a self-supervised framework to learn context-specific causal patterns from external causal statements. Then, we adopt a contrastive transfer strategy to incorporate the learned context-specific causal patterns into the target ECI model. Experimental results show that our method significantly outperforms previous methods on EventStoryLine and Causal-TimeBank (+2.0 and +3.4 points on F1 value respectively)., Accepted to Findings of ACL 2021

Background: To explore the epidemiology, clinical features, risk indicators, and long-term outcomes of neurological complications caused by veno-arterial extracorporeal membrane oxygenation (V-A ECMO).Methods: We retrospectively analyzed 60 adult patients who underwent V-A ECMO support in our unit from February 2012 to August 2020. These patients were separated into the neurological complications group (NC group) and the non-neurological complications group (nNC group). The differences in basic data and ECMO data between the two groups were compared. The data of long-term neurological prognosis were collected by telephone follow-up.Results: Thirty-nine patients (65.0%) had neurological complications. There were significant differences between the two groups in terms of median age, hypertension, median blood urea nitrogen, median troponin I (TNI), median lactic acid, pre-ECMO percutaneous coronary intervention, continuous renal replacement therapy (CRRT), median Sequential Organ Failure Assessment score, median Acute Physiology and Chronic Health Evaluation II score, median peak inspiratory pressure, median positive end expiratory pressure, and median fresh frozen plasma (P < 0.05). The median Intensive Care Unit length of stay (ICU LOS), 28-day mortality, median post-ECMO vasoactive inotropic score, non-pulsate perfusion (NP), and median ECMO duration of the NC group were significantly higher than those of the nNC group (P < 0.05). Furthermore, multiple logistic regression analysis revealed that TNI (P = 0.043), CRRT (P = 0.047), and continuous NP > 12 h (P = 0.043) were independent risk indicators for neurological complications in patients undergoing ECMO. Forty-four patients (73.3%) survived after discharge, and 38 patients (63.3%) had Cerebral Performance Category score of 1–2. And there were significant differences between the two groups in long-term neurological outcomes after discharge for 6 months (P < 0.05).Conclusion: The incidence of neurological complications was higher in patients undergoing V-A ECMO and was closely related to adverse outcomes (including ICU LOS and 28-day mortality). TNI, CRRT, and continuous NP > 12 h were independent risk indicators for predicting neurological complications in ECMO supporting patients. And the neurological complications of patients during ECMO support had significant adverse effect on long-term surviving and neurological outcomes of patients after discharge for 6 months.

Identifying causal relations of events is an important task in natural language processing area. However, the task is very challenging, because event causality is usually expressed in diverse forms that often lack explicit causal clues. Existing methods cannot handle well the problem, especially in the condition of lacking training data. Nonetheless, humans can make a correct judgement based on their background knowledge, including descriptive knowledge and relational knowledge. Inspired by it, we propose a novel Latent Structure Induction Network (LSIN) to incorporate the external structural knowledge into this task. Specifically, to make use of the descriptive knowledge, we devise a Descriptive Graph Induction module to obtain and encode the graph-structured descriptive knowledge. To leverage the relational knowledge, we propose a Relational Graph Induction module which is able to automatically learn a reasoning structure for event causality reasoning. Experimental results on two widely used datasets indicate that our approach significantly outperforms previous state-of-the-art methods.

Document-level relation extraction (RE) has attracted more and more attentions recently. Edge-oriented graph neural network (EoG) is a new neural network exhibiting greater potential than previous node-oriented graph neural networks for document-level RE. In this paper, we propose a novel EoG, called knowledge-aware edge-oriented GNN (KEoG) for document-level RE. In KEoG, we further introduce not only two types of nodes to represent documents and external knowledge respectively, but also soft F-Measure loss function to solve the inherent class imbalance problem in document-level RE. Experiments conducted on two document-level datasets show that KEoG outperforms other state-of-the-art methods for comparison on both intra-sentence and inter-sentence relation extractions, indicating that KEoG is an effective extension of EoG.

Background Family history information, including information on family members, side of the family of family members, living status of family members, and observations of family members, plays an important role in disease diagnosis and treatment. Family member information extraction aims to extract family history information from semistructured/unstructured text in electronic health records (EHRs), which is a challenging task regarding named entity recognition (NER) and relation extraction (RE), where named entities refer to family members, living status, and observations, and relations refer to relations between family members and living status, and relations between family members and observations. Objective This study aimed to introduce the system we developed for the 2019 n2c2/OHNLP track on family history extraction, which can jointly extract entities and relations about family history information from clinical text. Methods We proposed a novel graph-based model with biaffine attention for family history extraction from clinical text. In this model, we first designed a graph to represent family history information, that is, representing NER and RE regarding family history in a unified way, and then introduced a biaffine attention mechanism to extract family history information in clinical text. Convolution neural network (CNN)-Bidirectional Long Short Term Memory network (BiLSTM) and Bidirectional Encoder Representation from Transformers (BERT) were used to encode the input sentence, and a biaffine classifier was used to extract family history information. In addition, we developed a postprocessing module to adjust the results. A system based on the proposed method was developed for the 2019 n2c2/OHNLP shared task track on family history information extraction. Results Our system ranked first in the challenge, and the F1 scores of the best system on the NER subtask and RE subtask were 0.8745 and 0.6810, respectively. After the challenge, we further fine tuned the parameters and improved the F1 scores of the two subtasks to 0.8823 and 0.7048, respectively. Conclusions The experimental results showed that the system based on the proposed method can extract family history information from clinical text effectively.

In this study, the content characteristics, comprehensive pollution assessment, and morphological distribution characteristics of heavy metals (Mn, Cd, Cr, Pb, Ni, Zn, and Cu) were researched based...

BACKGROUND The discovery of antineutrophil cytoplasm antibody (ANCA) makes the early diagnosis of primary vasculitis possible, and also has important guiding significance for the diagnosis and treatment of secondary vasculitis. This study aimed to investigate the clinical significance of ANCA. MATERIAL AND METHODS ANCA was detected by indirect immunofluorescence assay (IIF), and anti-myeloperoxidase (MPO) antibody, and anti-proteinase 3 (PR3) antibody were detected by ELISA. The results were analyzed retrospectively. RESULTS Among 118 730 patients, a total of 5853 (4.93%) were positive for ANCA. In the positive cases, 3.98% were male and 6.33% were female, with significant differences (χ²=123.38, P

Event extraction, which aims to identify event triggers of pre-defined event types and their arguments of specific roles, is a challenging task in NLP. Most traditional approaches formulate this task as classification problems, with event types or argument roles taken as golden labels. Such approaches fail to model rich interactions among event types and arguments of different roles, and cannot generalize to new types or roles. This work proposes a new paradigm that formulates event extraction as multi-turn question answering. Our approach, MQAEE, casts the extraction task into a series of reading comprehension problems, by which it extracts triggers and arguments successively from a given sentence. A history answer embedding strategy is further adopted to model question answering history in the multi-turn process. By this new formulation, MQAEE makes full use of dependency among arguments and event types, and generalizes well to new types with new argument roles. Empirical results on ACE 2005 shows that MQAEE outperforms current state-of-the-art, pushing the final F1 of argument extraction to 53.4% (+2.0%). And it also has a good generalization ability, achieving competitive performance on 13 new event types even if trained only with a few samples of them.

Machine reading comprehension (MRC) has achieved significant progress on the open domain in recent years, mainly due to large-scale pre-trained language models. However, it performs much worse in specific domains such as the medical field due to the lack of extensive training data and professional structural knowledge neglect. As an effort, we first collect a large scale medical multi-choice question dataset (more than 21k instances) for the National Licensed Pharmacist Examination in China. It is a challenging medical examination with a passing rate of less than 14.2% in 2018. Then we propose a novel reading comprehension model KMQA, which can fully exploit the structural medical knowledge (i.e., medical knowledge graph) and the reference medical plain text (i.e., text snippets retrieved from reference books). The experimental results indicate that the KMQA outperforms existing competitive models with a large margin and passes the exam with 61.8% accuracy rate on the test set.

This paper introduces DuEE, a new dataset for Chinese event extraction (EE) in real-world scenarios. DuEE has several advantages over previous EE datasets. (1) Scale: DuEE consists of 19,640 events categorized into 65 event types, along with 41,520 event arguments mapped to 121 argument roles, which, to our knowledge, is the largest Chinese EE dataset so far. (2) Quality: All the data is human annotated with crowdsourced review, ensuring that the annotation accuracy is higher than 95%. (3) Reality: The schema covers trending topics from Baidu Search and the data is collected from news on Baijiahao. The task is also close to real-world scenarios, e.g., a single instance is allowed to contain multiple events, different event arguments are allowed to share the same argument role, and an argument is allowed to play different roles. To advance the research on Chinese EE, we release DuEE as well as a baseline system to the community. We also organize a shared competition on the basis of DuEE, which has attracted 1,206 participants. We analyze the results of top performing systems and hope to shed light on further improvements.

Besides expanding the spectrum of KCNQ1 mutations causing Jervell and Lange-Nielsen Syndrome (JLNS), the results showed diversity of its phenotypes, and emphasized the importance of molecular genetic analysis in confirming clinical diagnosis and making diagnosis possible before the emergency symptoms for deaf individuals.This study aimed to investigate four patients from three Chinese families with congenital hearing loss clinically and genetically.Genetic analysis of previously reported deafness genes based on massively parallel sequencing was conducted in more than five thousand Chinese hearing loss patients. Detailed clinical features of the patients with compound heterozygous or homozygous mutations of KCNQ1 gene were collected and analyzed.Compound mutations of KCNQ1 were found to be the genetic etiology of four patients from three families. Among the six KCNQ1 mutations, c.546C A was identified as a novel mutation, c.965C T had been reported in JLNS, while c.683 + 5G A, c.1484_1485delCT, c.905C T and c.1831G A were previously reported in LQT1. In addition to congenital profound hearing loss in all subjects, two sibling subjects showed typical JLNS cardiac phenotype of prolonged QTc and recurrent syncopal episodes. One subject presented not only JLNS, but also iron-deficiency anemia and epilepsy. The other subject did not present any cardiac phenotype.

Six groundwater samples were collected from four aquifers in Renlou coal mine, northern Anhui Province, China, and concentration, temporal variation and speciation of five heavy metals (Cr, Cu, Ni, Pb and Zn) were analyzed. The results of physicochemical parameters and major ions showed that the selected groundwater were representative samples of Quaternary aquifer (QA), coal measure aquifer (CA), Taiyuan limestone aquifer (TA) and Ordovician limestone aquifer (OA). Cr, Ni and Zn increased in the four aquifers from 2002 to 2014, but Pb decreased from 2006 to 2014. Cu concentration increased in QA and TA, but decreased in CA and OA. Although all monitoring results met the standards recommended by WHO and GB5749-2006, Cr and Cu in some groundwater samples exceeded GB/T 14848-93, especially in CA. Soluble metallic hydroxides were the dominant species of heavy metals in QA, while free ions were the main species in TA and OA. Excepted Ni2+ and Zn2+, CuCO3(aq) and PbCO3(aq) were the dominant species of Cu and Pb in CA, respectively, indicating more CO2 had been flooded and dissolved in groundwater from ground atmospheric environment. The enhanced concentration and special dominant species of heavy metals revealed that the ‘closed’ characteristics of CA in Renlou coal mine might have been broken due to coal mining activity.

Groundwater is important for either resource usage or safety of coal mining in north China. In this study, concentrations of eight kinds of heavy metals (Pb, Cd, Cu, Ni, Cr, Zn, Fe and Mn) in groundwater from three deep aquifers in the Wolonghu coal mine, northern Anhui Province, China have been analyzed for water quality assessment and water source identification. The results suggest that the groundwater have different concentrations of heavy metals among aquifers, which might be the results of different occurrence forms of heavy metals (e.g. Pb, Cd, Cu and Cr were adsorbed by iron hydroxides) and different kinds of water rock interactions (e.g. Mn and Zn originated from carbonate rocks). In comparison with the groundwater quality standard of China and WHO, most of them cannot be used for drinking directly but must be treated before drinking, especially the Pb, Cd and Fe contents. Moreover, hydraulic connection between aquifers has been identified by plots of factor scores and cluster analysis, which is similar to the results obtained by previous studies. Moreover, discriminant analysis demonstrated that heavy metals can be used for identification the source of inrush water in coal mines.

An all-fiber tunable laser was fabricated based on an acousto-optic tunable filter and a tapered fiber. The structure was of a high signal-to-noise ratio, therefore, no extra gain flattening was needed in the laser. In the experiment, the wavelength of the laser could be tuned from 1532.1 nm to 1570.4 nm with a 3-dB bandwidth of about 0.2 nm. Given enough nonlinearity in the laser cavity, it could also generate a sliding-frequency pulse train. The laser gains advantages of fast tuning and agility in pulse generation, and its simple structure is low cost for practical applications.

The present study was conducted to investigate the concentrations of heavy metals (HMs) in deep groundwater from coal mining area, including cadmium (Cd), chromium (Cr), copper (Cu), zinc (Zn), lead (Pb), and nickel (Ni). The samples were collected from different aquifers in four coal mines of northern Anhui province, China, which were unconsolidated formation (UF), coal measure aquifer (CA), Taiyuan limestone aquifer (TA), and Ordovician limestone aquifer (OA), respectively. HM concentrations from the four different sources were analyzed by atomic absorption spectrometer, and were found in the order of Zn>Ni>Pb>Cu>Cd>Cr (in UF), Ni>Zn>Pb>Cu>Cr>Cd (in CA), Ni>Zn>Pb>Cu>Cd>Cr (in TA), and Zn>Ni>Pb>Cu>Cr>Cd (in OA), respectively. Concentrations of Cu, Zn, and Cr were found within the quality guidelines set by Bureau of Quality and Technical Supervision of China (GB/T 14848-93) and World Health Organization, while the concentrations of Cd, Pb, and Ni were higher than their respective permissible limits. The enhanced concentrations of Ni, together with Cr, were considered to be affected by anthropogenic sources, since they both had high variable coefficient. Moreover, the inter-dependence of HMs and their pollution sources were further discussed using statistical techniques, including one-factor analysis of variance, Pearson correlation analysis, and principle component analysis.

Background Pulmonary edema is an important cause of complications and death in severe drowning. Continuous veno-venous hemofiltration (CVVH) may reduce pulmonary edema and thus may be a treatment modality for severe sea water drowning resuscitation. Methos 20 dogs were anesthetized and tracheally intubated. 10 ml/kg of sea water was infused into trachea in a minute. All animals developed signs of respiratory distress and severe hypoxia (PaO2

There is little information available about the hydrochemical characteristics of deep groundwater in the Linhuan coal-mining district, Northern Anhui Province, China. In this study, we report information about the physicochemical parameters, major ions, and heavy metals of 17 groundwater samples that were collected from the coal-bearing aquifer. The results show that the concentrations of total dissolved solids, electrical conductivity, and potassium and sodium (K(+) + Na(+)) in most of the groundwater samples exceeded the guidelines of the World Health Organization (WHO) and the Chinese National Standards for Drinking Water Quality (GB 5749-2006). The groundwater from the coal-bearing aquifer was dominated by the HCO3·Cl-K + Na and HCO3·SO4-K + Na types. Analysis with a Gibbs plot suggested that the major ion chemistry of the groundwater was primarily controlled by weathering of rocks and that the coal-bearing aquifer in the Linhuan coal-mining district was a relatively closed system. K(+) and Na(+) originated from halite and silicate weathering reactions, while Ca(2+) and Mg(2+) originated from the dissolution of calcite, dolomite, and gypsum or anhydrite. Ion exchange reactions also had an influence on the formation of major ions in groundwater. The concentrations of selected heavy metals decreased in the order Mn > Zn > Cr > Cu > Ni > Pb. In general, the heavy metal concentrations were low; however, the Cr, Mn, and Ni concentrations in some of the groundwater samples exceeded the standards outlined by the WHO, the GB 5749-2006, and the Chinese National Standards for Groundwater (GB/T 14848-93). Analysis by various indices (% Na, SAR, and EC), a USSL diagram, and a Wilcox diagram showed that both the salinity and alkalinity of the groundwater were high, such that the groundwater could not be used for irrigating agricultural land without treatment. These results will be significant for water resource exploiting and utilization in coal mine area.

An in-fiber Mach-Zehnder interferometer was proposed and fabricated, which was based on a sandwich-like etched single mode fiber driven by only one acoustic transducer. It succeeded the feature of fast tuning and would not introduce frequency shift in the transmission spectrum. Based on it, a fast tuning dual-wavelength laser with a two-wavelength spacing around 3.5 nm was proved with a tuning range of about 3.6 nm, covering wavelengths from 1561.6 nm to 1568.9 nm.

An extended theory of the mutually modulated cross-gain modulation was developed to make it valid under considerable modulated wavenumber-interaction length product. Parametric analysis of the extended theory shows that, in the modulated probe beam out of the gain medium, the spectral sensitivity of the first harmonic wave is associated with its amplitude. The larger the sensitivity is, the smaller the amplitude will be. An effective method was provided to monitor and achieve high spectral sensitivity at a pre-selected frequency detuning in practice. The validity of the extended theory was confirmed by a sensing experiment in a piece of ∼10-km fiber with 10 kHz modulation frequency, in which a fast light propagation with a sensitivity of 0.592 mrad/kHz was achieved.

Postinfectious acute glomerulonephritis (PIGN) may occur after various bacterial and viral infections. Hepatitis B virus (HBV) infection is a cause of chronic glomerulonephritis. We report here 10 cases (ages 7–20 years-old) of chronic HBV carriers with acute glomerulonephritis, with positive glomerular staining of hepatitis B surface antigen, and detectable presence of HBV DNA in the glomeruli. This form of PIGN, HBV-PIGN, has not been previously identified. To further characterize clinical and pathological features of HBV- PIGN, we selected 10 cases of age-matched non-HBV PIGN for comparison. While both HBV associated PIGN and non-HBV PIGN similarly presented as proteinuria, hematuria, and hypertension, there was a trend of higher acute kidney injury and worsened prognosis in HBV-PIGN. 6 months after the onset, 4 patients with HBV associated PIGN did not show improvement from the disease, whereas all patients with non-HBV PIGN had complete or partial recovery. Pathologically, both HBV associated PIGN and non-HBV PIGN showed typical diffuse glomerular endocapillary proliferation, but HBV associated PIGN differed from classical PIGN with much fewer sub-epithelial glomerular “hump-shape” immune complex depositions. In conclusion, we have identified a novel association of HBV infection with acute glomerulonephritis.

We provide an optical heterodyne detection configuration based on an all-fiber acousto-optic structure, which acts as both frequency shifter and coupler simultaneously, with a resolution of 0.01 nm from 10 kHz to 90 kHz.

An all-fiber tunable ring laser consisted of an acousto-optic tunable notch filter and a tapered fiber was demonstrated with a tunable wavelength range from 1532.1 nm to 1570.4 nm and a linewidth of 0.2 nm.

Based on the conversion between the fundamental mode (LP01) and the higher-order mode (LP11) in a tapered fiber via a whispering gallery mode resonator, an add/drop filter was proposed and demonstrated experimentally, in which the resonator only interacted with one tapered fiber, rather than two tapered fibers as in conventional configurations. The filter gains advantages of easy alignment and low scattering loss over the other filters based on tapered fiber and resonator, and will be useful in application.