基于BiLSTM-CRF的中医文本命名实体识别＊

中医药文本命名实体识别在中医药文本挖掘中占有重要地位，本文通过BiLSTM-CRF方法实现对中医医案文本进行命名实体识别，不仅实现了基本命名实体识别，通过对数据集按照中草药、疾病和症状三个类别进行标记，还能够进行命名实体类别识别。对中医药相关医案进行规整的10292条句子进行序列标注，基于word2vec的向量构建，从而进行模型训练迭代，得到了准确率为97.23%，召回率为89.47%，F值为88.34%的中医药命名实体识别模型。各类别识别中，中草药类别识别精准率为94.41%，召回率为94.36%，F值为94.38%；疾病类别精准率为80.92%，召回率为80.92%，F值为80.92%；症状类别精准率为75.68%，召回率为81.68%，F值为78.56%，人工测试模型效果较好，能够对医案数据进行实体识别。命名实体识别模型较多，但用于中医药相关命名实体识别模型数量微乎其微，构建中医药相关命名实体识别模型，将更加有效的推动中医药文本挖掘发展。     

基于文本挖掘技术的脑卒中文献计量分析：以PubMed数据库为例

目的 回溯并勾勒2010年-2020年全球脑卒中领域研究的主题演化历程与时空分布规律。方法 从PubMed数据库中获取2010年-2020年脑卒中相关文献，利用R 4.1.2的pubmedR软件包批量抓取文献基本信息，通过quanteda软件包构建语料库，使用结构主体模型进行分析，并运用ggplot2软件包进行国家或地区合作网络分析。结果 全球脑卒中研究2010年-2013年为快速增长期，增长率为11.94%～17.60%;2014年-2018年为增长平缓期，增长率为3%～4%左右；2019年-2020年再次快速增长，增长率为12.81%、17.96%。中国大陆地区在2012年、2020年增长率分别为102.99%、43.85%,在全球两个阶段的高速增长期中均做出了较大贡献。基于文献摘要文本，共归纳出11个主题类别，其中“临床研究”与“卒中康复”占较大比重，合计热度值为46.66%。我国近年研究热点主要集中在“分子遗传病学”与“动物实验”。结论 全球脑卒中研究热度持续增长，主题类别逐渐丰富。我国需进一步加强脑卒中前端预防和后端康复研究，优化防治策略。     

中药治疗类风湿性关节炎、痛风及骨性关节炎用药规律研究

本研究以中国生物医学文献数据库为载体，类风湿性关节炎、痛风及骨性关节炎为研究对象，采用数学和统计的方法，获取高频率、协同出现的关键药对，探寻中医治疗这3 种疾病的用药规律。结果显示，当归、黄芪、桂枝、芍药等为中药治疗类风湿性关节炎核心药物。牛膝、茯苓、黄柏、大黄等为治疗痛风核心药物。当归、独活、牛膝、白芍等为治疗骨性关节炎核心药。分析结果可知中医在治疗类风湿性关节炎、痛风及骨性关节炎时，大都采用活血化瘀的治疗方法。此外，在治疗类风湿性关节炎和骨性关节炎时多采用祛风湿法，在治疗痛风时多采用利水湿法。     

Automatic classification of radiological reports for clinical care

Radiological reporting generates a large amount of free-text clinical narratives, a potentially valuable source of information for improving clinical care and supporting research. The use of automatic techniques to analyze such reports is necessary to make their content effectively available to radiologists in an aggregated form. In this paper we focus on the classification of chest computed tomography reports according to a classification schema proposed for this task by radiologists of the Italian hospital ASST Spedali Civili di Brescia. The proposed system is built exploiting a training data set containing reports annotated by radiologists. Each report is classified according to the schema developed by radiologists and textual evidences are marked in the report. The annotations are then used to train different machine learning based classifiers. We present in this paper a method based on a cascade of classifiers which make use of a set of syntactic and semantic features. The resulting system is a novel hierarchical classification system for the given task, that we have experimentally evaluated.     

基于文本挖掘技术分析类风湿性关节炎、强直性脊柱炎、溃疡性结肠炎和哮喘中医用药规律

方剂可以概括为中医辨证论治体系的数据集合，文本挖掘技术可将隐含在数据中的配伍规律以可理解方式进行表述。本研究选择中国生物医学文献数据库为载体，类风湿性关节炎（RA）、强直性脊柱炎（AS）、溃疡性结肠炎（UC）和哮喘（Asthma）4 种自身免疫性疾病为研究对象，采用数学和统计的方法，获取高频率、协同出现的关键药对，探寻中医治疗这4 种疾病的用药规律。结果表明，采用这种计算方法总结出的4 种疾病常用中药规律与其病机是相符的，且黄芪作为这4 种疾病共同使用的常用中药，可能作用于自身免疫性疾病的特异性病理靶标。     

学术论文洗稿原理、测试及学术期刊技术性防范对策

【目的】高效识别洗稿学术论文,防范此类学术不端行为,提高科技期刊办刊质量。【方法】分析洗稿原理,收集洗稿测试数据,比较样本原文和洗稿成品的异同;利用人工智能服务平台和中国知网网站,有针对性地探索出基于关键词提取和比对的技术性方法。【结果】按提出的方法测试数个样本后,发现该方法能较准确地识别出洗稿学术论文。【结论】提出的识别洗稿学术论文的方法具初步有效性。待学术期刊编辑人员广泛测试,验证该方法的有效性后,可将该方法软件化或平台化,以利编辑人员日常使用。     

Psychological impact on dispatched local lay rescuers performing bystander cardiopulmonary resuscitation

AimWe studied the short-term psychological impact and post-traumatic stress disorder (PTSD)-related symptoms in lay rescuers performing cardiopulmonary resuscitation (CPR) after a text message (TM)-alert for out-of-hospital-cardiac arrest, and assessed which factors contribute to a higher level of PTSD-related symptoms.MethodsThe lay rescuers received a TM-alert and simultaneously an email with a link to an online questionnaire. We analyzed all questionnaires from February 2013 until October 2014 measuring the short-term psychological impact. We interviewed by telephone all first arriving lay rescuers performing bystander CPR and assessed PTSD-related symptoms with the Impact of Event Scale (IES) 4–6 weeks after the resuscitation. IES-scores 0–8 reflected no stress, 9–25 mild, 26–43 moderate, and 44–75 severe stress. A score ≥26 indicated PTSD symptomatology.ResultsOf all alerted lay rescuers, 6572 completed the online questionnaire. Of these, 1955 responded to the alert and 507 assisted in the resuscitation. We interviewed 203 first arriving rescuers of whom 189 completed the IES. Of these, 41% perceived no/mild short-term impact, 46% bearable impact and 13% severe impact. On the IES, 81% scored no stress and 19% scored mild stress. None scored moderate or severe stress. Using a multivariable logistic regression model we identified three factors with an independent impact on mild stress level: no automated external defibrillator connected by the lay rescuer, severe short-term impact, and no (very) positive experience.ConclusionLay rescuers alerted by text messages, do not show PTSD-related symptoms 4–6 weeks after performing bystander CPR, even if they perceive severe short-term psychological impact.