论文部分内容阅读
根据北京市2010~2014年药品抽验中收集到的269份不合格报告,运用关联规则分析中的Apriori算法挖掘数据中潜在的有价值信息。分析药品不合格项目与药品类型、剂型、被抽样单位类别、生产企业所在区域等属性之间的关联关系,对药品抽验工作提出建设性意见。根据预设的支持度和置信度阈值,得到20条“有趣”强关联规则。其中,药品崩解时限不合格与化学药品、片剂、针对性抽验的关联,以及药品重量差异不合格与中成药、片剂、东城区药检所针对性抽验的关联是直观难以发现的复杂关联关系。研究表明,运用数据挖掘算法分析药品抽验数据,可以有效发现新的规则和未能直观展示的知识,这种技术的应用将有助于更合理地分配药品抽验的工作资源,集中监测具有潜在高风险的药品类别,更大程度地保障公众用药安全。
According to 269 unqualified reports collected in drug sampling from 2010 to 2014 in Beijing, the Apriori algorithm in association rules analysis was used to mine the potentially valuable information in the data. Analysis of drug unqualified items and the type of drug, dosage forms, sample units were sampled, the region where the manufacturer and other attributes of the relationship between the drug sampling work put forward constructive comments. According to the default support and confidence threshold, get 20 “interesting” strong association rules. Among them, the unqualified drug disintegration time limit associated with chemical drugs, tablets, targeted sampling, as well as drug weight differences and proprietary Chinese medicines, tablets, Dongcheng District, drug-specific sampling association is intuitive difficult to find complex associations relationship. The research shows that the use of data mining algorithms to analyze drug sampling data can effectively find new rules and knowledge that can not be displayed visually. The application of this technique will help to allocate the working resources of drug sampling more reasonably and focus on the monitoring of drugs with potential high Risk of drug categories, to a greater extent to protect public safety of drugs.