摘要
目的探索空间插值法结合自回归移动平均模型(ARIMA)和灰色模型(GM模型)在人类免疫缺陷病毒(HIV)发病率的时间序列数据时空分布和预测分析中的应用与比较。方法从中国疾病预防控制中心传染病数据库中提取2007—2018年我国HIV发病数据,使用克里格空间插值法分析HIV平均发病率分布和时空变化趋势,构建ARIMA和GM(1,1)预测模型,通过平均绝对百分误差(MAPE)评价不同模型的预测效果。结果2007—2018年全国HIV发病率呈上升趋势,每年第2季度上升达到最高点,第3、4季度开始下降,第1季度下降达最低点。全国西南部和西北部HIV发病率从2007年开始持续升高(新疆、云南最高),2012年以后升高速度变化显著;东部、东北部以及中部部分地区HIV发病率相对较低(河北、山东最低),但也呈逐年上升趋势。ARIMA和GM(1,1)预测模型的MAPE值分别为2.5%和3.8%。结论2007—2018年全国HIV发病率呈上升趋势,具有一定的季节特征,西南省份平均发病率较高,北方省份发病率较低,12年期间平均发病率自西南向东北方向、西北向东南方向降低;ARIMA模型预测HIV发病率的效果稍优于GM(1,1)模型。
Objective To explore the application of spatial interpolation method combined with autoregressive integrated moving average(ARIMA)model and grey model(GM)in the prediction and analysis of spatiotemporal distribution of HIV incidence time series.Methods Data of HIV incidence in China from 2007 to 2018 were extracted from the infectious disease database of Chinese Center for Disease Control and Prevention,and the distribution and spatiotemporal trends of the mean HIV incidence were analyzed using the kriging interpolation method.Then,the ARIMA and GM(1,1)prediction models were constructed,and the prediction effects of different models were assessed by the Mean Absolute Percentage Error(MAPE).Results From 2007 to 2018,the incidence of HIV infection in China showed an increasing trend.The incidence peaked in the second quarter,declined in the third and fourth quarters,and reached the lowest level in the first quarter.In southwest and northwest China,HIV incidence increased continuously from 2007(Xinjiang and Yunnan had the highest rates),and significantly after 2012.The prevalence of HIV infection was relatively low in eastern,northeastern and parts of central China(Hebei and Shandong had the lowest rates),but also showed an increasing trend year by year.The MAPE values of ARIMA and GM(1,1)models for predicting HIV morbidity were 2.5%and 3.8%,respectively.Conclusion From 2007 to 2018,HIV incidence in China showed an increasing trend with prominent seasonal characteristics.The average incidence was higher in southwestern provinces and lower in northern provinces.Furthermore,the average incidence decreased from southwest to northeast and from northwest to southeast.The ARIMA model is superior to GM(1,1)model for predicting HIV morbidity.
作者
任聃
邱朔
杨鹏
尚峰
REN Dan;QIU Shuo;YANG Peng;SHANG Feng(Health Medicine Department,the Second Medical Center,Chinese PLA General Hospital,Beijing 100071,China;Global Health Office,Beijing Center for Disease Prevention and Control,Beijing 100013,China;Department of Health Statistics,School of Public Health,Air Force Medical University,Xi’an 710032,China;Medical Service Center,the 908^(th)Hospital of Chinese PLA Joint Logistic Support Force,Nanchang 330002,China)
出处
《南昌大学学报(医学版)》
2022年第6期71-75,81,共6页
Journal of Nanchang University:Medical Sciences
基金
全军后勤科技青年培育计划(20QNPY047)
军事医学提升计划项目(2020SWAQ10)。
关键词
人类免疫缺陷病毒
发病率
预测
空间插值法
自回归移动平均模型
灰色模型
时间序列
human immunodeficiency virus
incidence
prediction
spatial interpolation method
autoregressive integrated moving average model
grey model
time series