基于科研文献挖掘的疾病与药物关联研究

发布时间：2018-03-24 01:23

本文选题：科研文献挖掘　切入点：生物医学实体　出处：《山西医科大学》2017年硕士论文

【摘要】：目的:生物医学实体是包含在医学科研文献中出现的疾病、药物、基因等名称、术语或概念,即一种文献内含知识,了解其相互关联对于科学研究意义重大。然此类知识被大量淹没于文献海洋,亟需一种有效知识管理方式将之快速地展现给科研人员。鉴于此,本研究拟基于科研文献挖掘开展疾病与药物实体关联研究。方法:1.文献分析法通过搜集、鉴别、整理相关文献,分析当前国内外相关研究历史、现状及存在的问题。在阅读、整理、归纳、分析这些文献材料的基础上,借鉴他人的研究成果,从而形成自己的研究框架。2.编程语言和数据库技术利用Java和Python等程序设计语言和数据库技术将上述PubMed ID相关的文献信息下载,进行批量数据有序化,并在MySQL数据库中分别建库。3.生物实体识别方法使用基于词典的匹配方法识别疾病实体与药物实体。4.信息计量学方法利用Python自编程序语言,基于信息计量学中的共现关系构建疾病与药物实体共现网络,并运用词频分析和共词分析进行疾病与药物实体关联分析。5.社会网络分析方法利用社会网络分析工具Pajek对共现网络进行宏观和微观层次的指标分析。微观指标层次进行中心度(点度中心度、接近中心度、中介中心度)等指标的对比分析。最后利用Gephi对共现网络进行可视化分析。结果与结论:本研究所使用的生物医学实体识别及关联发现方法能够帮助科研人员从大规模的生物医学文本中快速探测被隐藏的关联,具有良好的推广性,也同样适用于疾病-基因、基因-药物等其他生物医学实体之间的分析。
[Abstract]:Objective: biomedical entity is to include in the medical research literature of the disease, medication, gene name, term or concept, which is a document containing knowledge, understand the relationship for scientific research is of great significance. However, such knowledge is submerged in the ocean of literature, need an effective knowledge management mode will quickly show for researchers. In view of this, this study intends to carry out the research on related diseases and drug entity mining based on scientific literature. Methods: 1. literature analysis by collecting, collating relevant literature, identification, analysis of the current domestic and foreign related research history, current situation and existing problems. In the reading, sorting, summary, analysis of these documents on the reference of others' research achievements, to form their own research framework of.2. programming language and database technology using Java and Python programming language and database technology of the P UbMed ID related information download, batch data ordering, and in the MySQL database were database.3. biological entity recognition method by using Python programming language matching method to identify disease entity and entity.4. drug information measurement dictionary based on information in metrology co-occurrence relationship building disease and drug entity co-occurrence based on the network, and the use of the word frequency analysis and co word analysis of disease associated with drug entity analysis.5. social network analysis method of co-occurrence analysis tool Pajek network to analyze the macro and micro level indicators. Using social network micro index level center (degree centrality, closeness centrality, betweenness centrality) comparative analysis other indicators. Finally, using Gephi visualization analysis of co-occurrence network. Results and conclusion: biomedical and entity recognition used in this study The joint discovery method can help researchers quickly detect hidden associations from large-scale biomedical texts, and has good generalization. It also applies to the analysis of diseases, genes, genes, drugs and other biomedical entities.

【学位授予单位】：山西医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1;R-05

【参考文献】