Hierarchical spatial process models for estimating and predicting health effects

描述（由申请人提供）：本申请涉及广泛的挑战领域（15）转化科学和特定挑战主题：（15- tw -101）预测气候变化对健康影响的模型。我们建议开发贝叶斯分层统计方法和软件，以帮助空间分析人员建立健康结果与大气和气候预测因子之间的关系。我们提出了一个全面的建模框架，以适应不同来源和类型的时空数据。在目标1中，我们提出了一个统计建模框架，用于模拟暴露、气候和健康结果数据，该框架集成了点级空间不对齐数据和使用贝叶斯分层空间模型的支持回归变化的方法。我们确定了三种健康结果：哮喘住院，非黑色素瘤皮肤癌的发病率和食源性疾病沙门氏菌病。目标2使用称为“预测过程”的降维随机过程修改这些模型，使其适用于大型数据集。最后，在目标3中，我们承诺提供一套软件包，帮助将必要的空间数据库和显示组件与贝叶斯统计建模能力集成在一起，从而将我们的方法提供给比目前更广泛的健康和环境研究人员和管理人员。识别明显更有害的环境和气候相关因素将提高卫生研究人员、政策制定者和患者的理解和决策过程，从而对卫生保健系统和社会产生深远的有益影响。通过使研究人员避免使用经常揭示欺骗性故事的临时和定性方法，我们提出的统计方法可以在公共卫生研究中产生深远的有益影响，这些研究可能会触及社会意想不到的角落。

DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (15) Translational science and Specific Challenge Topic: (15-TW-101) Models to predict health effects of climate change. We propose to develop Bayesian hierarchical statistical methods and software that will help spatial analysts to establish relationships among health outcomes and atmospheric and climate predictors. We propose a comprehensive modeling framework to accommodate disparate sources and types of spatial-temporal data. In Aim 1 we propose a statistical modelling framework for modelling exposure, climate and health outcome data that integrates methods for point-level spatially mis- aligned data and change of support regression using Bayesian hierarchical spatial models. We identify three health outcomes: asthma hospitalizations, incidence of nonmelanoma skin cancer and a food borne disease salmonellosis. Aim 2 modifies adapts these models for use with large datasets using a dimension reduction stochastic process called the "predictive process". Finally, in Aim 3, we promise a suite of software packages that help integrate necessary spatial databases and display components with Bayesian statistical modeling ca- pability, thus delivering our methodology to a far broader audience of health and environmental researchers and administrators than is currently accessible. Identifying environmental and climate-related factors that are pronouncedly more detrimental will improve the understanding and decision making process of health researchers, policy makers and patients, thereby having far-reaching beneficial effects on the health care system and society. By redeeming the investigators from using ad-hoc and qualitative methods that often reveal deceptive stories, our proposed statistical methods can have far reaching beneficial effects in public health research that will potentially touch unexpected corners of society.