应用空间回归技术从全局和局部两水平上定量探讨影响广西流行性乙型脑炎发病的气象因素

目的探讨空间回归技术在筛选影响广西壮族自治区（简称广西）流行性乙型脑炎（简称乙脑）发病的气象因素中的价值，为广西乙脑的区域性生态预防提供科学依据。方法利用空间滞后模型（spatiallagmod—el，SLM）和地理加权回归分析（geographicalweightedregression，GWR）从全局和局部2个水平上探讨广西乙脑发病率与气象因素间的量化关系。结果空间滞后模型拟合结果表明：AIC=18．69、R^2=0．64、模型残差独立。GWR拟合结果表明：AIC=14．99、R^2=0．62、模型残差独立；气象因素对乙脑发病的影响效应具有空间变异性，不同地区的地理加权回归系数β和拟合优度R^2不同。结论空间回归技术在探讨具有空间自相关性和异质性的疾病数据的影响因素时，较经典回归分析效果好。广西年均相对湿度、年均日照、年均气压是影响乙脑发病的主要气象因素，且具有空间变异性。

Study on the relationship between Japanese encephalitis and meteorological factors in Guangxi, based on global & local spatial regression model

Objective To analyze the value of spatial regression models in screening meteorological factors which affect the occurrence of Japanese encephalitis in Guangxi Zhuang Autonomous Region, provide evidence for ecological pre- vention of Japanese encephalitis. Methods Spatial lag model （SLM） and geographical weighted regression （GWR） were used to investigate the spatial effects between meteorological factors and incidence of Japanese encephalitis in Guangxi. Re- suits From spatial lag model, AIC was 18.69, RE was 0. 64, and the residual had no correlation. However, in geograph- ical weighted regression model, AIC was 14. 99, RE was 0. 62, and the residual had no correlation. Moreover, the effect of specific meteorological factor had spatial heterogeneous in Guangxi. Conclusions The resuhs derived from spatial models are apparently superior to traditional regression models. Average annual relative humidity, average annual sunshine and av- erage annual atmospheric pressure with spatial heterogeneity are the main meteorological factors affecting the occurrence of Japanese encephalitis.