图书情报工作 ›› 2021, Vol. 65 ›› Issue (12): 122-129.DOI: 10.13266/j.issn.0252-3116.2021.12.012

• 研究论文 • 上一篇    下一篇

基于链路预测的协同药物组合推荐研究:面向疾病并发症诊疗

雷鸣1, 夏梦鸽1, 汪雪锋1, 刘佳2   

  1. 1. 北京理工大学管理与经济学院 北京 100081;
    2. 中国传媒大学媒体融合与传播国家重点实验室 北京 100024
  • 收稿日期:2020-07-31 修回日期:2021-02-11 出版日期:2021-06-20 发布日期:2021-07-03
  • 通讯作者: 汪雪锋(ORCID:0000-0002-4857-6944),教授,博士,通讯作者,E-mail:wxf5122@bit.edu.cn
  • 作者简介:雷鸣(ORCID:0000-0003-1746-4090),博士研究生;夏梦鸽(ORCID:0000-0003-0612-5659),硕士研究生;刘佳(ORCID:0000-0003-2627-933X),副教授,在站博士后。
  • 基金资助:
    本文系国家自然科学基金面上项目"生物医学领域潜在颠覆性技术识别方法研究"(项目编号:72074020)研究成果之一。

Research on Drug Combination Recommendation Based on Link Prediction for Concurrent Diseases Treatment

Lei Ming1, Xia Mengge1, Wang Xuefeng1, Liu Jia2   

  1. 1. School of Management and Economics, Beijing Institute of Technology, Beijing 100081;
    2. State Key Laboratory of Media Convergence and Communication, Communication University of China, Beijing 100024
  • Received:2020-07-31 Revised:2021-02-11 Online:2021-06-20 Published:2021-07-03

摘要: [目的/意义] 药物组合相比单一药物在临床治疗中存在多种优势,但药物数量的快速增长为药物组合筛选和推荐带来挑战,因此设计有效的预测方法为药物研发人员推荐更易产生协同作用的药物组合从而提高筛选效率具有重要意义。[方法/过程] 面向疾病并发症诊疗需求,基于链路预测构建协同药物组合推荐模型,首先利用SAO语义挖掘识别医学文献中的并发症信息,在此基础上利用医学数据库构建"疾病-药物-靶点"异质网络,并引入链路预测方法对网络进行药物作用机制的相似性评估,预测哪些药物组合更可能产生协同作用,进而依据预测结果针对某个疾病或某对并发症进行药物组合推荐。[结果/结论] 肠道疾病数据实证分析结果表明协同药物组合预测模型具有有效性和实用性。

关键词: SAO语义挖掘, 链路预测, 社会网络分析, 药物组合推荐, 疾病并发症

Abstract: [Purpose/significance] Compared with single drug, drug combination has many advantages in clinical treatment. But the growth of drug quantity brings difficulties to drug combination screening experiment. Therefore, it is of great significance to design an effective prediction method to recommend drug combination which is more likely to produce synergistic effect for pharmaceutical staff, so as to improve the screening efficiency.[Method/process] For the need of concurrent diseases treatment, proposed a drug combination recommendation model based on link prediction, and used the SAO semantic mining to identify the complications in medical literature. On this basis, we used the medical database to build the heterogeneous "disease-drug-target" network, and introduced link prediction to evaluate the similarity of drug action mechanism, and predicted which drug combinations were more likely to have synergistic effect. Based on the prediction results, recommended a combination of drugs for a certain disease or a pair of complications.[Result/conclusion] The empirical analysis of intestinal disease data verified the practicality and efficiency of the model.

Key words: SAO semantic mining, link prediction, social network analysis, drug combination recommendation, disease complications

中图分类号: