Spatial Model-based Ecological Study On Geographical Health Events

Human health is currently a key part of study of geography and environmental science. The dissertation focuses on application of geographic information system and spatial model to spatial pattern and risk factors of geographic health. The author not only summarized the traditional methods of spatial analysis and model, but also tackled some problems in study of geographic health of small probability by proposing three spatial models which was then applied to two examples of geographic health issue. The dissertation was composed of the following two parts:(1) Theoretical studyThe dissertation firstly summarized methods of spatial data analysis and basic theory of spatial models, and then proposed three spatial models targeted at problems in application to geographical health of small probability. The Hierarchical Bayesian model was proposed to address the problem of instability of variance due to small cases of morbidity or mortality. Four geographical detectors were proposed to assess the environmental risks of health: the risk detector indicates where the risk areas are; the factor detector identifies factors that are responsible for the risk; the ecological detector discloses relative importance between the factors; and the interaction detector reveals whether the risk factors interact or lead to disease independently. And finally generalized linear geostatistical model was proposed to guard in assessment of environmental risks of health against spuriously significant covariate effects which might result from ignoring the spatial correlation inherent in the data.(2) Practical applicationThe dissertation used the Hierarchical Bayesian model to map the spatial pattern of under-five mortality in Wenchuan at the township scale. Based on this map, we used geographical detectors (risk detector, factor detector, ecological detector, and interaction detector) to assess effects of both physical factors and social-demographic factors on the under-five mortality. It was found that three factors are responsible for child mortality: earthquake intensity, housing collapse, and slope.The generalized linear geostatistical model, or exactly speaking, spatial Gaussian linear model was employed to explore association between The Yunnan Sudden death Syndrome and environmental factors and compared to ordinary Gaussian linear model. It was found that only NDVI, as a proxy of the little white mushroom, had significant effect on sudden death although both NDVI and precipitation were indentified as significant factors by ordinary Gaussian linear model ignoring spatial correlation between death ratios. In addition, we did not find precipitation and geological factors, e.g. stratum and fault, had significant effect on sudden death.In conclusion, the main content was summarized and the problems for further studied in this field were presented at the end of this dissertation.