摘要
目的:探索双黄连注射剂不良反应(ADR)发生的流行病学特点。方法:通过系统检索1979~2015年文献,采集双黄连注射剂ADR详细个例报道,在此基础上应用Microsoft ACCESS构建数据库,进而以Clementine 12.0和Weka 3.7为数据挖掘平台,应用C5.0和CART决策树算法和贝叶斯网络等方法开展数据挖掘研究。结果:共检索符合纳入条件的双黄连注射剂ADR详细病例569例。贝叶斯网络分析显示ADR临床表现与ADR发生时间等因素存在相关性。决策树C5.0算法显示输液开始后0~5 min与36~40 min两个时间段内,ADR案例中ADR临床诊断为过敏性休克的案例占42.254%;输液开始后6~10 min时间段内,ADR临床诊断为过敏性休克的案例占25.275%。CART算法显示输液开始后0~10 min与31~40 min两个时间段内25岁以上的患者中,发生过敏性休克的ADR案例占这两个时间段内发生的总ADR案例的39.655%,高于其他时间段。结论:双黄连注射剂ADR流行病学特点可能与ADR发生时间、患者年龄等因素相关,尚需更大样本量的数据分析验证。
Objective： To explore the epidemiological characteristics of Shuanghuanglian injection adverse reactions. Methods： Analysis the collected detail Shuanghuanglian injection case reported from 1979 to 2015,apply Microsoft Access Database platform and Clementine 12. 0 and Weka 3. 7 platform to data mining which include decision tree algorithm（ C5. 0 and CART） and bayesian networks method. Results： 569 Shuanghuanglian injection adverse reaction cases were collected. The result of Bayesian network analysis showed that adverse drug reaction（ ADR） clinical manifestation and the ADRs occurred time were correlation factors. The result of C5. 0 decision tree algorithm showed that in the period of 0-5min and 36-40 min,the allergic shock patients occupy 42. 254%,in the period of 6-10 min time frame was anaphylactic shock cases accounted for 25. 275%. The result of CART decision tree algorithm showed that in the period of 0-10 min and31-40 min,the proportion of patients who older than 25 years old suffered allergic shock was 25. 275%. Conclusion： Epidomiologi-cal characteristics of Shuanghuanlian injection＇s ADR might be related to such factors as the ADR occurrence time,patient age,which still needed a larger sample size analysis of the data validation.
出处
《药物流行病学杂志》
CAS
2017年第1期19-26,共8页
Chinese Journal of Pharmacoepidemiology
基金
国家自然科学基金项目(编号:81473547)
关键词
双黄连注射剂
数据挖掘
不良反应
决策树
贝叶斯网络
Shuanghuanglian injection
Adverse drug reactions
Data mining
Decision tree
Bayesian network
