Application Of GIS Technology To Analyze Spatial Distribution Of Diabetes In Xinjiang

Objective: This study aims to explore the spatial distribution of diabetes with spatial autocorrelation and spatial regression method, aims to carry out theory evidence for the policy making on disease prevention and control, and to offer reference for the same study. Methods: Diabetes prevalence data about 2014 year from Xinjiang Disease Prevention and Control Center(CDC)were accumulated, and get the Xinjiang map of scale with 1:1000000, input the code, prevalence data, population. Translate excel data into SHP format data. Moran’s I, Moran’s I scatter plot, G statistics, spatial regression analysis method can be realized with software Arc GIS10.1 and Geo DaTM0.9.5i.spatial cluster map and diabetes prevalence distribution map can be realized using GIS technology. Results: 1. A total of 318704 diabetes cases were reported, with the prevalence rate of 142.74 per 10000 population. The result presents a strong spatial autocorrelation of diabetes prevalence rate in Xinjiang with Moran’s I and Moran scatter plot analysis. 2. Global Getis-Ord G statistics indicate that there were high occurrence of diabetes(Getis=0.1242, P<0.01). 3. Anselin’s local indication of spatial association(LISA) of diabetes prevalence in 2014 identified that 33 counties have significance, which is negative and statistically significant in the region were six counties, the county have lowest value of LISA among these region is Fuhai county. the local autocorrelation indicators of six region diabetes prevalence showed negative correlation, cluster types were "high-low", "low-high". Local Moran’s I values of 27 regions showed positive correlation, cluster types of these regions were "high-high", "low-low". 4. The results of local G statistics showed that, there were 49 regions’ local G statistics value showed positive result, in which maximum value 4.9832 was(Kuerle city), These areas formed gathering areas of high diabetes prevalence rate, called hot spots. The Z(Gi)scores of 21 counties have negative correlation, called cold spots area. 5. Results of spatial regression analysis showed, through spatial lag model, we discover that the per capita GDP value influence on the registration rate of diabetes. The more per capita GDP value of region, the higher registration rate is. Conclusions: 1. The global spatial autocorrelation showed that the distribution of diabetes prevalence rate of Xinjiang were positively correlated in 2014. 2. The distribution of diabetes prevalence in Xinjiang counties showed high and low clustering type. Preventive measures should be take to control risk factors in high clustering areas, and more attention should be paid to find out the advantage factors to control diabetes in low clustering areas, to provide guidance for the prevention work of diabetes. 3. The per capita GDP value influence on the registration rate of diabetes. The more per capita GDP of region, the higher registration rate is. The results provide useful informations on the prevention and control of diabetes.