基于地理信息系统的中国食管癌地理流行病学研究-流行病与卫生统计学专业论文.docxVIP

汕头大学医学院硕士研究生毕业论文 汕头大学医学院硕士研究生毕业论文 IV IV software by using IDW interpolation and zonal statistics. Normalized difference vegetation indexes (NDVI) of the 237 sampling areas were extracted from the NDVI maps provided by the same database as above and finished in ENVI 4.2 software. Spearman correlation analysis was used to analyze the relationship between EC mortality of these areas and the spatial environment factors. The 237 sampling areas were classified as 4 death ranks according to their EC mortality by cluster analysis. Nonparametric test was used to compare geographic and climatic factors between the 4 death ranks. In order to control multicollinearity, multiple regression analysis after factor analysis between esophageal cancer and geographic and climate factors were also carried out to confirm the risk factors for EC. All the statistic analyses were finished in SPSS 13.0. Results: The counties that have the highest morality of EC show significant aggregation. The results of the digital distribution maps of EC mortality, drought index (DI) and water-heat index (WHI) show that high EC mortality mostly occurred in areas with high DI and low WHI. Overlay analysis shows that there are clear differences of soil and vegetation types between high- and low-risk areas of EC, and mean SOCD of high EC mortality areas is lower than that of low EC mortality areas (t= -2.47, p less than 0.05). Kruskal-Wallis and Jonckheere-Terpstra tests show that annual mean precipitation, DI, WHI, mean wind speed of June (p less than 0.01 respectively) and annual mean temperature, annual mean highest/lowest temperature, mean wind speed of 4,6,8,12 months (p less than 0.05 respectively) of the four EC death ranks are different. Spearman correlation analysis shows weak negative correlation between precipitation, WHI, highest/lowest temperature and EC mortality (r=-0.233, -0.404, -0.143, -0.128 respectively; and p less than 0.001, 0.001, 0.05, 0.05 respectively), and weak positive