Data Integration and Advanced Statistical Modeling to Describe and Control Pediatric Pedestrian Injuries in The United States

 描述(申请人提供)：在美国，每年有1,000名儿童死于儿童的行人伤害，每年有51,000人受伤，5,300人住院。我们在这个项目中的目标是通过量化和表征美国儿童行人损伤的小区域时空风险，应用、翻译和传播用于流行病学和人口健康研究的大数据分析方法。我们实现这一目标的具体目标是：(1)创建一个全面的国家儿童行人伤害数据库，使用时间序列来描述和分析美国的儿童行人伤害 和回归分析方法。(2)在县和人口普查区层面量化儿童行人伤害风险，以确定高危地区，并评估国家安全路线到学校干预计划的预防效果。以及，(3)使用在线街道图像来识别与十字路口相关的道路、人行道和十字路口特征，在这些十字路口，行人经常受伤，尽管附近已经实施了安全的上学路线干预措施。这些目的是为了检验以下假设：1)可以使用桌面系统有效地创建、操作和查询大量信息丰富的儿科伤害健康数据集，(2)集成嵌套拉普拉斯近似是在大型时空数据集中识别儿科伤害高危区域的实用、可靠和可访问的方法，以及(3)街道图像审计是现场访问的可行替代方案，以确定高危地区的风险因素。在项目期结束时，我们将在网上发布材料，供卫生研究人员复制与健康相关的数据集的方法，并为我们的分析结果创建一个简单的用户友好界面和数据查询系统，供研究人员、政策制定者和其他感兴趣的各方使用，为当地的伤害预防和控制工作提供信息。通过将计算机科学的进展应用、演示和转化为国家大型儿童健康数据，该应用程序响应NIH大数据到知识(BD2K)研究优先事项，以“解决所有生物医学研究人员在发布、访问、管理、分析和整合各种数据类型的数据集方面面临的挑战”，以及NICHD优先支持“关于儿科创伤的研究，包括预防、治疗和康复”，并将评估、演示和传播尖端计算机科学和统计工具，以使用可应用于其他流行病学和公共卫生研究的方法来解决紧迫的儿童健康问题。

 DESCRIPTION (provided by applicant): Pediatric pedestrian injury kills 1,000 children every year in the United States, and results in 51,000 annual injuries and 5,300 hospitalizations. Our goal in this project is to apply, translate and disseminate large-data analytic methods for epidemiology and population health research by quantifying and characterizing the small- area spatiotemporal risk of pediatric pedestrian injury in the United States. Our specific aims to achieve this goal are: (1) Create a comprehensive national database of pediatric pedestrian injuries to describe and analyze pediatric pedestrian injury in the United States using time-series and regression methods. (2) Quantify pediatric pedestrian injury risk at the county and census-tract level to identify high-risk areas, and evaluate the preventative effect of the National Safe Routes to School intervention program. And, (3) use online street imagery to identify road, sidewalk and intersection characteristics associated with intersections where pedestrians are commonly injured despite Safe Routes to Schools Interventions having been implemented in the neighborhood. These aims are designed to test the hypotheses that 1) Large informative pediatric injury health data sets can be efficiently created, manipulated and queried using desktop systems, (2) Integrated nested Laplace approximations are a practical, reliable and accessible approach to identifying high-risk areas for pediatric injury in large spatiotemporal datasets, and (3) Street imagery audits are a feasible alternative to site visits to identify risk factors in high-risk areas. At the end of the project period, we will post online materials for heath researchers to replicate the methods for health-related data sets, and create a simple user-friendly interface and data query system for the results of our analyses that can be used by researchers, policy makers and other interested parties to inform local injury prevention and control efforts. By applying, demonstrating and translating advances in computer science for large national child health data the application is responsive to the NIH Big Data to Knowledge (BD2K) research priories to "address the challenges facing all biomedical researchers in releasing, accessing, managing, analyzing, and integrating datasets of diverse data types" and the NICHD priority to support "research on pediatric trauma, including prevention, treatment, and rehabilitation", and will evaluate, demonstrate and disseminate cutting edge computer science and statistical tools to address a pressing child health issue using approaches that can be applied to other epidemiological and public health research.