基于ARIMA模型对我国药品不良反应报告例数时间序列分析及预测  

Analysis and Prediction of the Time Series of Adverse Drug Reactions in China Based on an ARIMA Model

作　　者：林彦全 李明[2] 唐婉 吴刚 郑芳芳 陈红斗 郭伟 LIN Yanquan;LI Ming;TANG Wan;WU Gang;ZHENG Fangfang;CHEN Hongdou;GUO Wei(School of Pharmacy,Xuzhou Medical University,Xuzhou 221004,China;Department of Drug Monitoring,Jiangsu Adverse Drug Reaction Monitoring Center,Nanjing 210002,China;Department of Pharmacy,Affiliated Suqian Hospital of Xuzhou Medical University,Suqian 223800,China)

机构地区：[1]徐州医科大学药学院,徐州221004 [2]江苏省药品不良反应监测中心药品监测科,南京210002 [3]徐州医科大学附属宿迁医院药学部,宿迁223800

出　　处：《药学与临床研究》2023年第6期519-522,共4页Pharmaceutical and Clinical Research

基　　金：江苏省卫计委高层次卫生人才“六个一工程”拔尖人才科研项目(LGY2020071)。

摘　　要：目的:应用自回归移动平均(ARIMA)时间序列模型分析及预测我国药品不良反应(ADR)报告例数的发生状况及趋势,为监管及临床提供参考。方法:收集我国药品不良反应监测中心2009~2021年ADR报告的数量,运用SPSS 23.0软件对我国ADR报告的数量进行ARIMA模型的建模拟合,对ADR报告的时间序列进行分析。结果:根据2009~2021年每年ADR的数量为基础建立时间序列,构建的最优模型为ARIMA(0,2,0),所建模型R^(2)>0.8,拟合度较好,标准化贝叶斯信息准则(BIC)为5.564,平均绝对误差(MAE)为10.199,ADR报告数量呈逐年上升趋势,增长超3倍之多。结论:ADR报告例数可以运用ARIMA模型进行分析,我国ADR报告例数与ARIMA模型拟合的情况基本一致。Objective:To analyze and predict the incidence and trend of adverse drug reaction(ADR)reports in China by using autoregressive integrated moving average(ARIMA)time series model,so as to provide reference for supervision and clinical practice.Methods:The number of ADR reports in China Adverse Drug Reaction Monitoring Center from 2009 to 2021 was collected,and the ARIMA model was constructed by using SPSS23.0 software to simulate the number of ADR reports in China,and the time series of ADR reports were analyzed.Results:The time series was established based on the annual number of ADR from 2009 to 2021.The optimal model was ARIMA(0,2,0),and the established model was R^(2)>0.8,the fitting degree was good,the standardized Bayesian information criterion(BIC)was 5.564,and the mean absolute error(MAE)was 10.199.The number of ADR reports showed an increasing trend year by year,increasing by more than 3 times.Conclusion:The number of ADR cases can be analyzed by ARIMA model,and the number of ADR cases in China is basically consistent with the fitting situation of ARIMA model.

关 键 词：药物不良反应 时间序列分析 ARIMA模型 预测 

分 类 号：R969.3[医药卫生—药理学]