ARIMA乘积季节模型在山东省肺结核月报告发病率预测中的应用  

APPLICATION OF ARIMA MULTIPLE SEASONAL MODEL IN PREDICTING THE INCIDENCE OF TUBERCULOSIS IN MONTHLY REPORT

作　　者：梁士强 高丽洁 张子悦 杨亮 孙怡 LIANG Shi-qiang;GAO Li-jie;ZHANG Zi-yue(Feixian County Center for Disease Control and Prevention,Feixian,China 273400;Shandong Medical College,Linyi,China 276000)

机构地区：[1]费县疾病预防控制中心,山东费县273400 [2]山东医学高等专科学校,山东临沂276000

出　　处：《山东医学高等专科学校学报》2024年第4期17-20,共4页Journal of Shandong Medical College

摘　　要：目的应用专家建模器建立ARIMA乘积季节模型,分析其在肺结核月报告发病率预测中的适用性。方法应用SPSS专家建模器对2017年1月-2021年12月山东省肺结核月报告发病率资料自动建立ARIMA乘积季节模型,预测2022年1-12月发病率并与观测值比较,评估预测效果。结果ARIMA(1,0,0)(0,1,1)12为最简模型,该模型的R^(2)、MAPE、正态化BIC分别为0.588、6.820、-2.459;预测结果显示,预测波动趋势与实际一致,观测值均在预测值的95%可信区间范围内,相对误差1.79%~46.71%。结论应用专家建模器建立ARIMA乘积季节模型操作简单,适于对肺结核月报告发病率进行短期预测。Objective To establish the ARIMA multiple seasonal model with an expert modeler to analyze its applicability in predicting the incidence of tuberculosis in monthly report.Methods SPSS expert modeler was used to automatically establish an ARIMA multiple seasonal model with the incidence rate of tuberculosis in Shandong Province from January,2017 to December,2021 to predict the incidence rate from January to December of 2022 and to compare the result with the observed data to evaluate the prediction effect.Results ARIMA(1,0,0)(0,1,1)12 was the simplest model,and the R^(2),MAPE and normalized BIC of this model were 0.588,6.820 and-2.459 respectively.The results showed that the predicted fluctuation trend was consistent with the real data,and the observed values were all within the 95%confidence interval of the predicted value,with a relative error between 1.79%and 46.71%.Conclusion ARIMA multiple seasonal model is simple and suitable for short-term prediction of the monthly reported incidence of tuberculosis.

关 键 词：专家建模器 肺结核 ARIMA模型 预测 

分 类 号：R521[医药卫生—内科学]