基于配对检验的ARIMA模型在我国甲肝发病数预测中的应用  

Application of ARIMA model based on paired test in the prediction of hepatitis A incidence in China

作　　者：丁勇[1] 张蓓蓓[1] 吴静 DING Yong;ZHANG Beibei;WU Jing(Kangda College,Nanjing Medical University,Lianyungang 222000;School of Biomedical Engineering and Information,Nanjing Medical University,Nanjing 211166,China)

机构地区：[1]南京医科大学康达学院,江苏连云港222000 [2]南京医科大学生物医学工程与信息学院,江苏南京211166

出　　处：《南京医科大学学报（自然科学版）》2024年第10期1456-1461,共6页Journal of Nanjing Medical University(Natural Sciences)

基　　金：国家自然科学基金(61901225);江苏省高校自然科学研究(19KJD330001);江苏高校哲学社会科学研究(2022SJYB1868);江苏省大学生创新创业训练计划(202213980011Y);南京医科大学康达学院第一期青年教师科研导师项目(KD2022KYDS020);南京医科大学康达学院第二期品牌专业建设工程(JX206000302);南京医科大学康达学院医学信息模拟及预测科研团队(KD2022KYCXTD003)。

摘　　要：目的:探讨基于配对检验的求和自回归移动平均(autoregressive integrated moving average,ARIMA)模型在我国甲肝发病预测中的应用,提出时间序列模型预测效果评价的新思路与方法。方法:根据2004年1月—2021年12月我国甲肝传染病月发病数建立ARIMA模型,对2022年1—8月的甲肝月发病数进行预测,通过配对t检验和误差分析评估该模型的预测效果。结果:配对t检验结果显示,ARIMA(1,1,0)(0,1,1)12模型预测的甲肝月发病数与实际月发病数差异无统计学意义(P>0.05),说明模型有较好的预测能力,预测结果的相对误差平均值为3.86%,标准差为3.25%。结论:ARIMA乘积季节模型能够较准确地预测我国甲肝的发病趋势;配对检验为时间序列模型预测效果的评价提供了客观评价依据,较好地解决了时间序列模型预测效果的评价问题。Objective:To explore the application of autoregressive integrated moving average(ARIMA)model based on paired test in predicting the incidence of hepatitis A in China,and put forward a new idea and method for evaluating the prediction effect of time series model.Methods:An ARIMA model was established for the monthly incidence of hepatitis A infectious diseases in China from January 2004 to December 2021,and the monthly incidence of hepatitis A infectious diseases from January to August 2022 was predicted.The prediction effect of the model was evaluated by paired t-test and error analysis.Results:The results of paired t-test showed that there was no significant difference between the monthly incidence of hepatitis A predicted by ARIMA(1,1,0)(0,1,1)12 model and the actual monthly incidence of hepatitis A(P>0.05),indicating that the model had good prediction ability,and the mean relative error and standard deviation of the prediction results were 3.86%and 3.25%.Conclusion:ARIMA product season model can accurately predict the incidence trend of hepatitis A in China.The paired test provides an objective basis for evaluating the prediction effect of time series model,and solves the problem of evaluating the prediction effect of time series model well.

关 键 词：配对检验 甲型肝炎 ARIMA乘积季节模型 预测 

分 类 号：R512.6[医药卫生—内科学]