高原地区青海省流感病毒活动规律的时间序列分析及预测模型研究  

Time series analysis on activity law of influenza virus in plateau regions of Qinghai Province and study on prediction model

作　　者：赵金华[1] 龙江[2] 马永成[1] 石燕[1] 马斌忠[1] 曹海兰[1] 徐莉立[1] ZHAO Jin-hua;LONG Jiang;MA Yong-cheng;SHI Yan;MA Bin-zhong;CAO Hai-lan;XU Li-li(Qinghai Province Center for Disease Control and Prevention,Xining 810007,China;不详)

机构地区：[1]青海省疾病预防控制中心,西宁810007 [2]重庆市疾病预防控制中心

出　　处：《医学动物防制》2021年第11期1030-1034,共5页Journal of Medical Pest Control

基　　金：国家“十三五”科技重大专项传染病监测技术平台项目(2017ZX10103006-002);青海省应用基础研究和自然科学基金项目(2019-ZJ-7046)。

摘　　要：目的构建时间序列自回归移动平均模型(autoregressive integrated moving average, ARIMA)乘积季节性模型,预测高原地区青海省流感病毒的活动趋势,探讨该模型在预测流感病毒活动规律中的应用。方法对2009—2019年流感病毒核酸阳性检出率的时间序列资料构建模型,以流感病毒核酸阳性检出率为验证数据,验证预测模型效果。结果高原地区青海省2009—2019年流感病毒的活动呈现明显季节性效应,ARIMA(1,1,0)(0,1,0);为最优模型,回归系数差异有统计学意义(P<0.05),其中R^(2)=0.765,BIC=4.77,白噪声残差分析显示序列自相关函数Ljung-Box=19.080(P=0.324),残差为随机误差,实际值在预测值的95%可信区间范围内。结论乘积季节模型ARIMA(1,1,0)(0,1,0);较好地拟合和预测了短期内流感病毒的活跃程度,显示2020年青海省流感病毒发生暴发或流行趋势较低,可为全省流感流行和暴发起到预警作用。Objective To predict the activity law of influenza virus in plateau regions of Qinghai Province, by building the seasonal model of autoregressive integrated moving average(ARIMA) product of time series, and to discuss the application of this model in predicting the activity rule of influenza virus.Methods The model was built based on the data of time series for positive rate of influenza virus nucleic acids from 2009 to 2019,and the positive rate of influenza virus nucleic acid was taken as the validation data to verify the effects of the prediction model.Results The influenza virus activity in Plateau regions of Qinghai from 2009 to 2019 showed obvious seasonal effects, ARIMA(1,1,0)(0,1,0);was found to be the optimal model, and the difference in the regression coefficient was statistically significant(P<0.05),of which R^(2)=0.765 and BIC=4.77 held, white noise residuals analysis showed that sequence autocorrelation function Ljung-Box=19.080(P=0.324),the residual was the random error, the actual value was within the confidence interval in 95% of predicted value.Conclusion The product season model ARIMA(1,1,0)(0,1,0);fits well and predicts the activity of influenza virus in the short term, shows that there will be an outbreak or low epidemic trend of influenza virus in Qinghai Province in 2020,and it plays early warning role for influenza epidemics and outbreaks.

关 键 词：时间序列 自回归移动平均模型 流感病毒 活动规律 预测模型 

分 类 号：R373.13[医药卫生—病原生物学]