ARIMA季节性模型在河南省猩红热发病预测中的应用  

Application of ARIMA seasonal model in predicting the incidence of scarlet fever in Henan

作　　者：潘静静[1] 王文华 王莹莹 其木格[1] 申晓靖[1] 王海峰[1] 叶莹[1] 黄学勇[1] PAN Jingjing;WANG Wenhua;WANG Yingying;QI Muge;SHEN Xiaojing;Wang Haifeng;YE Ying;HUANG Xueyong(Institute for Infectious Disease Prevention and Control,Henan Provincial Center for Disease Control and Prevention,Zhengzhou,Henan450016,China)

机构地区：[1]河南省疾病预防控制中心传染病所,河南郑州450016

出　　处：《河南预防医学杂志》2022年第12期899-902,912,共5页Henan Journal of Preventive Medicine

摘　　要：目的 建立河南省猩红热发病的ARIMA季节模型,探讨该模型在预测猩红热发病中的应用。方法 以2005-2019年河南省猩红热病例数为基础建立ARIMA模型,并用2019年1-12月的实际发病数进行验证,利用最优模型预测2020年1-12月河南省猩红热发病数。结果 最优模型为ARIMA (2,1,2)(0,1,1)12,参数均有统计学意义(P<0.05),BIC=7.406,残差序列为白噪声(Ljung-Box=17.234,P=0.141),拟合效果较好。2019年1-12月猩红热月发病数预测值和实际值的平均相对误差为9.9%,实际值均在预测值95%置信区间(95%CI)。预测河南省2020年猩红热平均月发病数为192例,少于2019年的203例,明显多于2020年实际月均病例数47例。结论 ARIMA(2,1,2)(0,1,1)12模型拟合河南省猩红热发病数效果较好,可用于河南省猩红热发病趋势的短期预测。Objective To establish an ARIMA seasonal model of Scarlet fever incidence, and explore the feasibility of the model in predicting the incidence of scarlet fever in Henan province. Methods The ARIMA model was established based on the monthly cases of scarlet fever from 2005 to 2019 in Henan province. The cases from January to December in 2019 was used to test the prediction results of the model. Then the optimal model was used to predict the cases from January to December of 2020. Results The prediction model of scarlet fever was ARIMA(2,1,2)(0,1,1)12, and the parameters were all statistically significant(P<0.05). The minimum model goodness of fit value of BIC was 7.406,and the residual sequence was white noise(Ljung-Box=17.234,P=0.141).The goodness of fit was high. The Mean relative error between actual values and fitted values from January to December in 2019was 9.9%. The actual values were within the 95% confidence interval(95%CI) of the predicted values. The average monthly incidence of scarlet fever in Henan province in 2020 was predicted to be 192 cases, which was lower than that in 2019(203 cases),and distinctly higher than the average monthly actual cases in 2020(47 cases). Conclusion The model of ARIMA(2,1,2)(0,1,1)12can well simulate the trend of scarlet fever, and can be used for the short-term prediction of scarlet fever in Henan province.

关 键 词：ARIMA模型 猩红热 发病预测 模型应用 

分 类 号：R196[医药卫生—卫生事业管理]