Limited Mixed Model Of Count Data With Space Effect

Nowadays, AIDS has become a global problem, there are many governments realized the seriousness of the harm, wanton spread of AIDS is bound to affect the social-economic development and social harmony, but also on the country’s population development strategy and policy planning a huge impact influenced by many factors, drug is one of the important factors. Drug abuse is the main cause cumulative reported HIV infections in Yunnan Province ranks first in the country. Since 1989, for the first time in Ruili, Yunnan province found ways to HIV infection through injecting drug use cases since the AIDS epidemic continues to rise, the majority of those infected with unemployed youth, followed by workers and peasants. But in 2004--2011 years, the Yunnan HIV through injecting drug use decreased the number of cases overall. HIV infection among drug users in Yunnan Province, mainly for ethnic minorities and the western border of the southwestern border, showed significant spatial clustering feature.In the international context, the relationship between AIDS among drug users about the research has been 20,30 years of history. Early HIV epidemiology research is given priority to with qualitative analysis and simple descriptive statistics analysis, although in recent years on drug users for HIV/AIDS related studies have relied on the appropriate statistical model and spatial effect of the time people into account, but the current research work are mostly around completely it is necessary to develop simple Logistic regression model, for each region the sentinel test results in the complex count data, categorical data, nominal data and missing data involves less, in the domestic to international epidemiological studies in recent years the preferred way-"diseases, Mapping" technology is also little reported.In this paper, Yingjiang, Linxiang, Baoshan, Qujing, Zhaotong drug users living with HIV as the research object, collected the drug use, the number of sexual partners and other indicator data, Bayesian analysis from the perspective of the start, establish drug people living with HIV Hurdle model, using "Disease Mapping", considering the spatial random effects regression model, and establish a reasonable space auto regressive structure, to estimate the relative risk of spatial effects, assessing the danger areas and regions of Yunnan Province. According to different data classification AIDS pathogen, test different categories of potential differences in the return structure, and analyze the differences between the infected person’s own factors, lifestyle factors, and behavioral factors associated with drug among different pathogens; on this basis, establish the original data Bayesian finite mixture modeling framework, using DIC information criterion to identify potential category number, and make a comparison with the former.