我国血吸虫病报告病例数ARIMA模型预测研究  

Application of ARIMA model to predict schistosomiasis cases in China

作　　者：常雪莲[1] 王小莉[1] 魏星 李亮[3] Chang Xuelian;Wang Xiaoli;Wei Xing;Li Liang(Department of Microbiology and Parasitology,Bengbu Medical College,Anhui Key Laboratory of Infection and Immunity,Bengbu 233030,China;Anhui Key Laboratory of Digital Medicine and Intelligent Health,Bengbu Medical College,Bengbu 233030,China;Department of Food and Biological Engineering,Bengbu College,Bengbu 233000,China)

机构地区：[1]蚌埠医学院病原生物学教研室安徽省感染与免疫重点实验室,安徽蚌埠233030 [2]蚌埠医学院数字医学与智慧健康安徽省重点实验室,安徽蚌埠233030 [3]蚌埠学院食品与生物工程系,安徽蚌埠233000

出　　处：《中华地方病学杂志》2021年第9期712-717,共6页Chinese Journal of Endemiology

基　　金：安徽省高校自然科学研究重点项目(KJ2019A320、KJ2019A325、KJ2020A0566);蚌埠医学院自然科学研究项目(BYKY1819ZD、BYKY1825ZD)。

摘　　要：目的研究我国(不含港、澳、台地区)的血吸虫病月报告病例数进行自回归移动平均(autoregressive integrated moving average,ARIMA)模型预测作用,为血吸虫病的防控提供科学依据。方法采用ARIMA模型,以2009年1月至2018年12月我国血吸虫病月报告病例数时间序列为训练集,应用R 3.6.2软件进行平稳性分析后,采用赤池信息准则和贝叶斯信息准则等筛选参数,选出较优ARIMA模型;以2019年1-12月我国血吸虫病月报告病例数为测试集进行验证和逐月优化,得到1个最优ARIMA模型;并以2019年1月至2020年10月我国血吸虫病月报告病例数验证最优ARIMA模型的预测效果。结果基于2009年1月至2018年12月数据,可以得到4种较优ARIMA模型,分别为ARIMA(2,0,2)(1,0,1)[12]、ARIMA(2,0,2)(0,0,1)[12]、ARIMA(2,0,2)(1,0,0)[12]和ARIMA(2,0,2);以2019年1-12月的病例数实际值和4种ARIMA模型预测值分别进行对比,构建出的血吸虫病月报告病例数的最优预测模型为ARIMA(2,0,2)(1,0,1)[12];预测的相对误差均值为0.51%。结论本研究构建的ARIMA模型精度较高,适用于我国血吸虫病病例数的短期预测分析,可为该病防治提供数据支持,具有一定实践指导意义。Objective An autoregressive integrated moving average(ARIMA)model was used to predict the number of monthly reported cases of schistosomiasis in China(excluding Hong Kong,Macao and Taiwan),so as to provide a scientific basis for prevention and control of schistosomiasis.Methods Using ARIMA model,taking the time series of monthly reported cases of schistosomiasis in China from January 2009 to December 2018 as the training set,after stabilizing analysis with R 3.6.2 software,ARIMA models were selected by using screening parameters such as akaike information criterion and bayesian information criterion.Taking the number of monthly reported cases of schistosomiasis in China from January to December 2019 as the test set for verification and monthly optimization,an optimal ARIMA model was obtained.The prediction effect of the optimal ARIMA model was verified by the number of monthly reported cases of schistosomiasis in China from January 2019 to October 2020.Results Based on the data of monthly reported cases of schistosomiasis in China from January 2009 to December 2018,four ARIMA models were obtained,namely ARIMA(2,0,2)(1,0,1)[12],ARIMA(2,0,2)(0,0,1)[12],ARIMA(2,0,2)(1,0,0)[12]and ARIMA(2,0,2).By comparing the actual number of cases from January to December 2019 with the predicted values of the four ARIMA models,the optimal prediction model of monthly reported cases of schistosomiasis was ARIMA(2,0,2)(1,0,1)[12],and the mean relative error of the prediction was 0.51%.Conclusions The ARIMA model constructed in this study has high accuracy and is suitable for short-term prediction and analysis of the number of schistosomiasis cases in China.It can provide data support for prevention and control of the disease,and has certain practical guiding significance.

关 键 词：血吸虫病 月报告病例数 时间序列 ARIMA模型 预测 

分 类 号：R532.21[医药卫生—内科学]