期刊名称：Brazilian Journal of Probability and Statistics
印刷版ISSN：0103-0752
出版年度：2006
卷号：20
期号：1
页码：29-47
出版社：Brazilian Statistical Association
摘要：We model the number of cases of dengue fever in Rio deJaneiro, Brazil, between the 48t hepidemiological week of 2001 and the 20thepidemiological week of 2002. This period comprises the worst dengue feverepidemic the city has ever experienced. The original data were aggregatedacross time and classified according to the district where the infected per-son lived. We consider the number of cases of the disease for each district asfollowing a Poisson random variable. We compare among many di.erent mod-els that consider social-economical and infrastructural covariates plus randome.ects to explain the logarithm of the mean of this Poisson. Following theBayesian paradigm, we consider these random e.ects to follow a ConditionalAutoregressive (CAR) prior. Our main contribution resides on entertainingdi.erent neighborho od structures for these e.ects. These di.erent neighbor-hood structures take into account the particular landscape of the city of Riode Janeiro, which is known for its high mountains. We have observed somesignificant e.ect of social-economics and infra structural factors on the rela-tive risk of the disease. The great variety of results that we provide has givenevidence that the choice of the neighborhood structure when the CAR prioris used, might be of importance, as the significance of the covariates changes,when we use di.erent neighborhood structures
关键词：Conditional autoregressive models; geographic information;system; model comparison; neighborhood structure
