中国药物警戒 ›› 2017, Vol. 14 ›› Issue (6): 346-349.

• 统计学在药物警戒中的应用专栏 • 上一篇    下一篇

改进的加权关联规则算法在识别企业生产风险的应用

孙怡园1,邓剑雄2,杨悦1,*   

  1. 1 沈阳药科大学工商管理学院，辽宁 沈阳 110016；
    2 广东省不良反应监测中心，广东 广州 510080
  • 收稿日期:2017-05-05 修回日期:2017-08-17 出版日期:2017-06-20 发布日期:2017-08-17
  • 通讯作者: 杨悦，女，教授，药事管理。E-mail: yyue@vip.163.com。
  • 作者简介:孙怡园，女，在读硕士，数据统计分析与建模。

Application of Improved Weighted Association Rule Algorithm in Identifying the Risk of Pharmaceutical Production

Sun Yiyuan1, Deng Jianxiong2, Yang Yue1,*   

  1. 1 College of Business Administration, Shenyang Pharmaceutical University, Liaoning Shenyang,110016, China;
    2 Center for ADR Monitoring of Guangdong, Guangdong Guangzhou, 510080, China
  • Received:2017-05-05 Revised:2017-08-17 Online:2017-06-20 Published:2017-08-17

PDF (PC)

115

摘要： 目的 改进加权关联规则算法，建立企业风险监测模型。方法 在Apriori关联规则算法的理论基础上，以“肝损害”、“肾损害”等药害事件常见的器官损害类别作为关联规则项目，将严重占比、新的不良反应作为关联规则权重指标，进一步改进加权关联规则算法，并利用2006年1-6月广东省药品不良反应监测中心接收的ADR报告数据，识别“齐二药事件”风险，验证企业风险监测模型的可行性。结果 在最小支持度为0.05，最小置信度为0.90的情况下，改进的加权Apriori关联算法识别出“齐二药事件”风险，而传统Apriori关联算法无风险信号产生；降低最小支持度至0.02时，传统算法虽检测出“齐二药”风险信号，但出现了大量的混杂风险信号。结论 相对于传统加权关联算法，改进的加权关联模型针对性强、准确度高，更有利于企业生产风险的预警和监测。

关键词: Apriori关联规则, 加权关联规则, 风险预警模型, 不良反应监测

Abstract: Objective To improve the weighted association rule algorithm, and establish enterprise risk monitoring model. Methods Based on the theory of Apriori association rule algorithm, the category of organ damage, such as "liver damage" and "kidney damage", is used as the association rule item, and the serious events or the new adverse events are used as the weight index of association rules, in order to further improve the weighted association rule algorithm. Besides, we use the ADR report data received by the Guangdong Provincial Adverse Reaction Monitoring Center from January to June 2006 to identify the risk of "Qiqihar Second Pharmaceutical Factory Drug Safety Event" and verify the feasibility of the enterprise risk monitoring model. Results In the case of minimum support of 0.05 and minimum confidence of 0.90, the improved weighted Apriori correlation algorithm identifies the risk of " Qiqihar Second Pharmaceutical Factory Drug Safety Event ", while the traditional Apriori correlation algorithm has no risk signal generation; reduces the minimum support to 0.02 , although the traditional algorithm come up to risk signal, but there have been a lot of mixed risk signal as well. Conclusion Compared with the traditional weighted correlation algorithm, the improved weighted correlation model is highly effective and accurate, and is more conducive to the early warning and monitoring of production risk.

Key words: Apriori association rule, weighted association rule, risk early-warning model, adverse reaction monitoring.

中图分类号: