基于机器学习的针灸相关疾病、基因、药物新关联挖掘  

Machine learning-based mining of new associations in acupuncture-related diseases,genes and drugs

作　　者：魏星[1] 谢静[1] 蒋秀林[1] 叶枫[1] 张德成[1] 陈友春[1] 朱文婕[1] WEI Xing;XIE Jing;JIANG Xiu-lin;YE Feng;ZHANG De-cheng;CHEN You-chun;ZHU Wen-jie(Bengbu Medical College Public Basic Courses School,Bengbu 233030,Anhui Province,China)

机构地区：[1]蚌埠医学院公共基础课学院

出　　处：《中华医学图书情报杂志》2019年第8期8-16,共9页Chinese Journal of Medical Library and Information Science

基　　金：安徽高校自然科学重点研究项目“疾病-基因-药物关系抽取关键技术与实证研究”(数字医学与智慧健康安徽省重点实验室KJ2019A0325);安徽省质量工程项目“计算机与程序设计”(2018mooc281);蚌埠医学院自然科学重点项目“基于词典和机器学习的基因实体识别机制研究”(BYKY1825ZD)

摘　　要：目的:挖掘针灸相关疾病、基因和药物间的新关联。方法:提出一种基于SVM的机器学习算法,结合词典识别疾病、基因和药物实体并挖掘三者之间的关联,构建针灸相关疾病、基因和药物关联网络。结果:识别出针灸相关的296种疾病、51种基因和278种药物,并在27种疾病、13种基因和135种药物之间挖掘出704种关联,构建3个关联网络,发现了262种新关联。结论:针灸相关疾病-基因-药物之间存在大量程度不一的关联,为针灸精准医疗提供了新的研究思路。Objective To mine the new associations in acupuncture-related diseases,genes and drugs.Methods A support vector machine(SVM)-based machine learning algorithm was proposed and different association networks for acupuncture-related diseases,genes and drugs were established by identifying the diseases,genes and drugs with dictionaries and mining their associations.Results A total of 296 acupuncture-related diseases,51 genes and 278 drugs were identified,and 704 associations were mined in 27 diseases,13 genes and 135 drugs.Three association networks were established,which discovered a total of 262 new associations in acupuncture-related diseases,genes and drugs.Conclusion New associations are detected in acupuncture-related diseases,genes and drugs,which can thus provide certain new ideas for studying the precision treatment of diseases by acupuncture.

关 键 词：针灸 疾病 基因 药物 SVM 关联网络 数据挖掘 文本挖掘 

分 类 号：R245[医药卫生—针灸推拿学] TP181[医药卫生—中医临床基础]