Spatial Generalized Linear Mixed Models With Multivariate CAR Models For Areal Data

This paper is based on CAR model which is commonly used in this type of problem and combines it with the generalized linear model.From the past,it only studied single diseases,in this paper.we developed two more diseases in space.Jin et al.introduced a more flexible generalized multivariate CAR model in 2005 by using Bayesian methods for parameter estimation.In this paper,we use the hierarchical Bayes method to calculate each parameter based on the study of Jin et al.In Sections 1 and 2,we introduce the theory of spatial autocorrelation,because it’s necessary if we want to do further research.Then we review the existing multivariate CAR model and extend it to space.Linear mixed model.In the third quarter,the simulation work will be done.The simulation test is performed on the mixed model of the MCAR model and the generalized linear model.It is proved that the model is reasonable and could be used for case analysis.In section 4,the paper use the data of the incidence and death of malignant tumors in 25 cancer registration areas in 2008,China.Moran’s I index is used to prove that the tumor incidence and death data are spatially related.According to Local Moran’s I value and Z score.the type of spatial aggregation in each region is determined,respectively H&H(abbreviations of high&high),H&L,L&h,L&L.Among them,taking Liaoning Province as an example for special analysis.In this paper,we selected stomach cancer and liver cancer for analysis,using Openbugs software to make all major parameters reach a stable state after 50,000 iterations,the results can be concluded that both diseases have a strong spatial correlation,and between the two disease,there has a weak positive relationship in space,stomach cancer has a positive promoting effect on liver cancer in space.