药物不良反应知识库的建设以及统计分析

目的:建立一个药物与不良反应对应关系的知识库,为药物不良反应的监测以及药物不良反应研究提供一个在线资源。方法:首先通过网络爬虫从互联网获取中文的药品说明书数据,然后通过自然语言处理的方法自动从药品说明书中提取相应的不良反应,并构建相应的知识库来存储药品与不良反应的关系,最后对该知识库药品以及药品的分类进行了不良反应的统计分析。结果:建立了一个由3 683个药品和39 583个药品与不良反应关系组成的知识库,统计表明76.6%的药物有明确不良反应,其中94.5%的药物都有多于1个的不良反应;给出了不良反应最多的10种药品以及最常见的不良反应;通过对不良反应最多的100个药物的ATC分类统计显示其中多数属于抗感染药物、心血管药物和抗肿瘤药及免疫制剂三类。结论:搭建了该药物不良反应知识库网站提供免费在线查询和下载http://www.cktp.org:8006/Display/ADE,可以服务于临床药物不良反应监测和相关的不良反应研究。

Construction and statistical analysis of an adverse drug reaction knowledgebase

OBJECTIVE To construct a knowledgebase including the relation between drugs and adverse reactions and to provide an online resource for the study of adverse drug reactions (ADRs) and pharmacovigilance.METHODS Firstly we collected drug labels from the Internet by web Crawlers. Then the corresponding adverse reactions of each drug were extracted automatically from drug labels using natural language processing technology. Finally we designed an appropriate knowledgebase to store the relationship between drugs and adverse reactions. Furthermore, we conduct a statistical analysis of adverse reactions based on drugs and drug classes in the knowledgebase.RESULTS We constructed a knowledgebase composed of 3 683 drugs and 39 583 relationships between drugs and adverse reactions. Statistics indicated that 76.6% drugs had evident adverse reactions, in which more than 94.5% had more than one adverse reaction. The top ten drugs with most adverse reactions and the top common adverse reactions were listed respectively. The ATC classification results of the top 100 drugs with the most adverse reactions showed that anti-infective drugs, cardiovascular drugs and antineoplastic and immunomodulating agents were the largest classifications with adverse drug reactions.CONCLUSION This study provides an online knowledge resource for ADR study and pharmacovigilance. This ADR knowledgebase can be accessed and downloaded at http://www.cktp.org:8006/Display/ADE.