Character-level neural network for biomedical named entity recognition

Biomedical named entity recognition (BNER), which extracts important named entities such as genes and proteins, is a challenging task in automated systems that mine knowledge in biomedical texts. The previous state-of-the-art systems required large amounts of task-specific knowledge in the form of feature engineering, lexicons and data pre-processing to achieve high performance. In this paper, we introduce a novel neural network architecture that benefits from both word- and character-level representations automatically, by using a combination of bidirectional long short-term memory (LSTM) and conditional random field (CRF) eliminating the need for most feature engineering tasks. We evaluate our system on two datasets: JNLPBA corpus and the BioCreAtIvE II Gene Mention (GM) corpus. We obtained state-of-the-art performance by outperforming the previous systems. To the best of our knowledge, we are the first to investigate the combination of deep neural networks, CRF, word embeddings and character-level representation in recognizing biomedical named entities.     

Sand Fly Fauna (Diptera,Pcychodidae, Phlebotominae) in Different Leishmaniasis-Endemic Areas of Ecuador,Surveyed Using a Newly Named Mini-Shannon Trap

To study the sand fly fauna, surveys were performed at four different leishmaniasis-endemic sites in Ecuador from February 2013 to April 2014. A modified and simplified version of the conventional Shannon trap was named “mini-Shannon trap” and put to multiple uses at the different study sites in limited, forested and narrow spaces. The mini-Shannon, CDC light trap and protected human landing method were employed for sand fly collection. The species identification of sand flies was performed mainly based on the morphology of spermathecae and cibarium, after dissection of fresh samples. In this study, therefore, only female samples were used for analysis. A total of 1,480 female sand flies belonging to 25 Lutzomyia species were collected. The number of female sand flies collected was 417 (28.2%) using the mini-Shannon trap, 259 (17.5%) using the CDC light trap and 804 (54.3%) by human landing. The total number of sand flies per trap collected by the different methods was markedly affected by the study site, probably because of the various composition of species at each locality. Furthermore, as an additional study, the attraction of sand flies to mini-Shannon traps powered with LED white-light and LED black-light was investigated preliminarily, together with the CDC light trap and human landing. As a result, a total of 426 sand flies of nine Lutzomyia species, including seven man-biting and two non-biting species, were collected during three capture trials in May and June 2014 in an area endemic for leishmaniasis (La Ventura). The black-light proved relatively superior to the white-light with regard to capture numbers, but no significant statistical difference was observed between the two traps.     

The role of reactive oxygen species in arsenite and monomethylarsonous acid-induced signal transduction in human bladder cells: acute studies

Arsenicals are known to induce ROS, which can lead to DNA damage, oxidative stress, and carcinogenesis. A human urothelial cell line, UROtsa, was used to study the effects of arsenicals on the human bladder. Arsenite [As(III)] and monomethylarsonous acid [MMA(III)] induce oxidative stress in UROtsa cells after exposure to concentrations as low as 1 microM and 50 nM, respectively. Previous research has implicated ROS as signaling molecules in the MAPK signaling pathway. As(III) and MMA(III) have been shown to increase phosphorylation of key proteins in the MAPK signaling cascade downstream of ErbB2. Both Src phosphorylation (p-Src) and cyclooxygenase-2 (COX-2) are induced after exposure to 50 nM MMA(III) and 1 microM As(III). These data suggest that ROS production is a plausible mechanism for the signaling alterations seen in UROtsa cells after acute arsenical exposure. To determine importance of ROS in the MAPK cascade and its downstream induction of p-Src and COX-2, specific ROS antioxidants (both enzymatic and non-enzymatic) were used concomitantly with arsenicals. COX-2 protein and mRNA was shown to be much more influenced by altering the levels of ROS in cells, particularly after MMA(III) treatment. The antioxidant enzyme superoxide dismutase (SOD) effectively blocked both As(III)-and MMA(III)- associated COX-2 induction. The generation of ROS and subsequent altered signaling did lead to changes in protein levels of SOD, which were detected after treatment with either 1 microM As(III) or 50 nM MMA(III). These data suggest that the generation of ROS by arsenicals may be a mechanism leading to the altered cellular signaling seen after low-level arsenical exposure.     

清肝益气降压方配伍针刺对高血压大鼠血清一氧化氮、血浆内皮素的影响

目的:观察清肝益气降压方配伍针刺太冲、太溪、足三里对高血压大鼠血压、血浆内皮素、血清一氧化氮的影响,探讨方药和针刺不同穴位的交互作用及其内在的机理。方法:采用正交设计安排实验,将造模成功大鼠分别给予清肝益气降压方和针刺,连续治疗2周后,测定大鼠血压、血浆内皮素、血清一氧化氮。结果:清肝益气降压方配伍针刺能使ET含量降低,NO含量升高,使血压降低,优于单纯针刺和方药组(P<0.05)。结论:针刺和药物间产生协同作用,可以按照中医辨证施治原则配伍使用。针刺和方药对各项指标的影响程度不一,可能这种多靶点、多途径、多层次的作用方式,是针药产生协同作用的基本原因。     

情志顺势团体辅导对大学生网络依赖的干预效果研究

目的 依据中医学顺势利导、扶正祛邪、心身合一等治疗原则,制定适用于网络依赖的团体心理辅导方案,引导网络依赖成员宣泄心理能量,协助成员建立对网络全面客观的评价,发现心中正向的认知及情感,最终摆脱网络依赖.方法 以某中医药大学26名网络依赖大学生作为实验组,26名同质学生作为对照组,对其网络依赖行为进行团体辅导干预.在团体辅导前、结束及结束后6周时施测修订后的中文网络成瘾量表(CIAS-R).结果 团体辅导前实验组与对照组CIAS-R得分无显著差异,在辅导结束(F=11.29,P＜0.01)及结束后6周(F=16.89,P＜0.01),实验组网络使用得分显著低于对照组;实验组内干预前、后及6周追踪测试CIAS得分有显著差异(F=11.29,P＜0.01),对照组内3时间段CIAS得分无显著差异.结论 情志顺势团体辅导对大学生网络依赖有明显的干预效果.     

文本挖掘在生物医学领域中的应用及其系统工具

系统介绍了生物医学文本挖掘的具体流程和文本挖掘技术在生物医学领域中的应用情况,并着重从自然语言处理和本体、命名实体识别、关系抽取、文本分类与聚类、共现分析、系统工具及评价、可视化等方面分别做了阐述.     

基于实体词典与机器学习的基因命名实体识别

将实体词典以特征的形式引入到机器学习模型中,提出一种基于实体词典与机器学习的基因命名实体识别方法,在GENIA 3.02语料上进行实验。测试结果表明引入实体词典特征后,在获得较高实体识别准确率的同时,优化CRFs识别模型的时间复杂度,提高系统识别效率。     

中文电子病历命名实体识别方法研究

目的/意义 探索基于中文电子病历的命名实体识别方法在构建医学知识图谱和相关应用推广方面的技术可行性。方法/过程 采用真实医疗电子病历数据对词嵌入表示模型进行精化,构建医学术语专有嵌入表示,并利用卷积神经网络等多模型提取局部语义特征,实现基于堆叠注意网络的中文医疗命名实体识别。结果/结论 堆叠注意网络模型F1值达到91.5%,较其他模型具备更强的医疗命名实体识别性能。进一步解决中文医疗命名实体识别难点,在实现全局语义特征全面深入提取的同时降低时间成本。     

Stacked ensemble combined with fuzzy matching for biomedical named entity recognition of diseases

Biomedical Named Entity Recognition (Bio-NER) is the crucial initial step in the information extraction process and a majorly focused research area in biomedical text mining. In the past years, several models and methodologies have been proposed for the recognition of semantic types related to gene, protein, chemical, drug and other biological relevant named entities. In this paper, we implemented a stacked ensemble approach combined with fuzzy matching for biomedical named entity recognition of disease names. The underlying concept of stacked generalization is to combine the outputs of base-level classifiers using a second-level meta-classifier in an ensemble. We used Conditional Random Field (CRF) as the underlying classification method that makes use of a diverse set of features, mostly based on domain specific, and are orthographic and morphologically relevant. In addition, we used fuzzy string matching to tag rare disease names from our in-house disease dictionary. For fuzzy matching, we incorporated two best fuzzy search algorithms Rabin Karp and Tuned Boyer Moore. Our proposed approach shows promised result of 94.66%, 89.12%, 84.10%, and 76.71% of F-measure while on evaluating training and testing set of both NCBI disease and BioCreative V CDR Corpora.     

基于中文医药文本的实体识别和图谱构建

知识图谱技术促进了新药研发的进展,但国内研究起点晚且领域知识多以文本形式存储,图谱重用率低。因此,本研究基于多源异构的医药文本,设计了以Bert-wwm-ext预训练模型为基础,并融合级联思想的中文命名实体识别模型,从而减少了传统单次分类的复杂度,进一步提高了文本识别的效率。实验结果显示,该模型在自建的训练语料上的F1分数达0.903,精确率达89.2%,召回率达91.5%。同时,将模型应用于公开数据集CCKS2019上,结果显示该模型能够更好地识别中文文本中的医疗实体。最后,利用此模型构建了一个中文医药知识图谱,图谱包含13 530个实体,10 939个属性,以及39 247个相关关系。本研究所提出的中文医药实体识别与图谱构建方法,有望助力研究者加快医药知识新发现,从而缩短新药研发进程。