A Spatial-Temporal Modleing Approach for Environmental Epidemiological Data

描述（由申请人提供）：需要收集不同时间和不同地理区域的环境流行病学数据。需要对这些数据进行分析，以确定国家环境政策的重要方面，即保护公民健康和防止对基础设施和环境造成损害的方面。本研究的目的是开发一个统计框架和方法的综合分析的时空数据的空气污染浓度和其他环境因素，暴露，健康结果和协变量信息。通常，这些不同的数据层在时间上是不对齐的，并且在不同的空间尺度上被观察到。本研究的重点是： [1]制定新的统计方法和模型，以调查环境压力因素之间的时空关联，同时考虑到人类活动，并结合两个案例研究对人类健康产生不利影响：* 研究臭氧和PM的影响（精细，过程和超细）对心血管死亡率在整个美国的连续研究臭氧和PM的影响（细、中、超细）在威斯康星州对哮喘、心脑血管疾病的疗效。 [2]建立一个广泛的统计框架，研究环境因素与不良健康结果之间的关系。该框架综合考虑了环境过程的参数依赖和非参数依赖结构，并考虑了空间错位、支持度的时空变化、时空协方差函数的平稳性和可分性的缺乏。暴露模拟器模型用于描述人群暴露水平。 [3]多变量时空环境流行病学数据的模型拟合、估计与预测。 [4]对确定性和随机模型的性能进行统计评估，以及模型诊断。在目标2-4中，我们建立了一般统计框架，这些框架将被应用到目标1中介绍的案例研究中。

DESCRIPTION (provided by applicant): Environmental epidemiological data need to be collected over time and across different geographic domains. These data need to be analyzed in order to determine important aspects of national environmental policy, aspects that protect the health of citizens and prevent damage to infrastructure and the environment. The purpose of this research is to develop a statistical framework and methodology for integrated analyses of spatial temporal data on air pollution concentrations and other environmental agents, exposure, health outcomes and covariate information. Generally, these various data layers are temporally misaligned and are observed at different spatial scales. The focus of this research is: [1] the development of new statistical methods and models for the investigation of the spatial and temporal association between environmental stressors, taking into account human activity, and adverse human health outcomes in the context of two case studies: *study of the impact of ozone and PM (fine, course and ultrafine) on cardiovascular mortality across the conterminuous U.S. *study of the impact of ozone and PM (fine, course and ultra fine) on asthma, cardiovascular and cerebrovascular diseases in the state of Wisconsin. [2] The development of a broad statistical framework to study the association of environmental factors and adverse health outcomes. This general framework incorporates parametric and nonparametric ial dependence structure for environmental processes, taking into account spatial misalignment, spatial and temporal change of support, and lack of stationarity and lack of separability in the space-time covariance function. An exposure simulator model is used to characterize population exposure levels. [3] The model fitting, estimation and prediction of multivariate space-time environmental epidemiological data. [4] The statistical assessment of the performance of deterministic and stochastic models, and model diagnostics. In aims 2-4 we establish general statistical frameworks that will be implemented to the case studies introduced in aim 1.