摘要
目的研究2011-2016年全国31个省、直辖市和自治区的客运量、人均国内生产总值(gross do mestic product,GDP)、人口密度和每千人医疗机构床位数对艾滋病发病数影响的时空变化特性,为预防艾滋病提供依据。方法建立时空加权泊松回归模型,采用局部线性地理加权回归方法和迭代加权最小二乘估计对系数函数进行估计及可视化,分析不同地区、不同年份下宏观因素对艾滋病发病数影响的时空非平稳性。结果全国各地区艾滋病发病区存在明显的时空聚集性和变化趋势;不同地区、不同时间的宏观因素对艾滋病发病数的影响各不相同。结论拟合优度诊断统计量(R^2,AIC,MSE)验证时空加权泊松回归模型拟合效果优于泊松回归模型,更好地反映时空数据中时空交互效应和非平稳特征,表明中国艾滋病发病数的时空分布与四个宏观因素的变化密切相关。
Objective In this paper,passenger quantity,GDP per capita,population density and the number of beds per thousand were investigated to reflect the spatio-temporal trend of AIDS/HIV incidence of 31 provinces,municipalities and autonomous regions from 2011 to 2016 in China,in order to provide a reference for controlling the spread of AIDS/HIV. Methods A geographically and temporally weighted Poisson regression model was established. The coefficient function was estimated and visualized according to locally linear geographical weighted regression method and iterative weighted least square estimation. Some spatio-temporal non-stationary properties of AIDS/HIV cases in different time and regions were studied. Result There existed the temporal and spatial characteristics and trend in the high-incidence areas. Macro factors with different times and regions had different influences on the number of AIDS/HIV cases. Conclusion Statistics for goodness of fit (R^2 ,AIC,MSE) showed GTWPR model was better than Poisson regression model,which could reflect spatio-temporal interaction and non-stationary characteristics. Result showed that the spatial and temporal distribution of AIDS/HIV epidemic was closely related to four macro factors.
作者
孙舒曼
李智明
张辉国
胡锡健
SUN Shu-man;LI Zhi-ming;ZHANG Hui-guo;HU Xi-jian(College of Mathematics and System Science,Xinjiang University,Urumqi 830046,China)
出处
《中华疾病控制杂志》
CAS
CSCD
北大核心
2018年第12期1207-1210,1215,共5页
Chinese Journal of Disease Control & Prevention
基金
国家自然科学基金(11661076)
国家社会科学基金(16BTJ024)
新疆维吾尔自治区科技计划项目(2016D01C043)
新疆维吾尔自治区高校科研计划项目(XJEDU2017M001)
关键词
艾滋病
宏观因素
时空加权泊松回归模型
时空非平稳性
Acquired immunodeficiency syndrome
Macro factors
Geographically and temporally weighted Pois-son regression model
Spatio-temporal non-stationary
