DEVELOPMENT OF DATA ONTOLOGIES FOR INTEGRATING MULTI-CENTER CARDIOVASCULAR STUDIE

心血管疾病及其相关风险因素，如高血压和血脂异常，由于死亡率和发病率的增加以及卫生保健费用的增加，构成了一个主要的公共卫生负担。需要大量的流行病学数据来检测许多个体基因和环境对这些性状的微小影响。然而，只有通过整合多个流行病学研究，才能得出强有力的推论。对来自多个研究的信息进行有意义的整合需要数据本体的发展，这使得以最佳方式整合研究间的信息成为可能，从而最大化信息内容，从而提高检测小效应大小的统计能力。数据集成的第二个复杂问题是，管理此类研究数据的软件应用程序通常是不可互操作的，即数据的“筒仓”，并且不能以语法和语义上有意义的方式共享。因此，需要一个以可互操作的方式整合所有研究的基础设施，以确保流行病学心血管研究在目前正在进行的生物医学信息学革命中仍然是可行的和主要的参与者。癌症生物医学信息网格（caBIGTM）通过开发能够交换信息或通过访问使用受控词汇表进行语义注释的元数据在语法上可互操作的软件系统来解决癌症领域中的这些问题。我们的首要目标是开发用于整合来自多个研究的心血管流行病学数据的本体。具体来说，我们提出了三个目标：首先，为三个不同的多中心流行病学研究开发心血管数据本体和词汇表，促进跨研究的数据整合和对各种表型的数据挖掘。其次，采用一种技术基础设施，利用模型驱动体系结构（MDA）和caBIGTM工具，利用心血管数据本体和词汇表，促进心血管数据集的集成和广泛共享。第三，主要通过与已建立的心血管研究网格（CVRG）合作，促进跨临床、转化和流行病学领域的研究社区之间的无缝数据共享和广泛数据传播。

Cardiovascular disease (CVD) and its associated risk factors such as hypertension and dyslipidemia constitute a major public-health burden due to increased mortality and morbidity and rising health care costs. Massive epidemiological data are needed to detect the small effects of many individual genes and the environment on these traits. However, sample sizes needed to make powerful inferences may only be reached by integrating multiple epidemiological studies. Meaningful integration of information from multiple studies requires the development of data ontologies which make it possible to integrate information across studies in an optimum manner so as to maximize the information content and hence the statistical power for detecting small effect sizes. A second compounding problem of data integration is that software applications that manage such study data are typically non-interoperable, i.e. “silos” of data, and are incapable of being shared in a syntactically and semantically meaningful manner. Consequently, an infrastructure that integrates across studies in an interoperable manner is needed to ensure that epidemiological cardiovascular research remains a viable and major player in the biomedical informatics revolution which is currently underway. The cancer Biomedical Informatics Grid (caBIGTM) is addressing these problems in the cancer domain by developing software systems that are able to exchange information or that are syntactically interoperable by accessing metadata that is semantically annotated using controlled vocabularies. Our overarching goal is to develop ontologies for integrating cardiovascular epidemiological data from multiple studies. Specifically, we propose three Aims: First, develop cardiovascular data ontologies and vocabularies for each of three disparate multi-center epidemiological studies that facilitate data integration across the studies and data mining for various phenotypes. Second, adopt a technology infrastructure that leverages the cardiovascular data ontologies and vocabularies using Model Driven Architecture (MDA) and caBIGTM tools to facilitate the integration and widespread sharing of cardiovascular data sets. Third, facilitate seamless data sharing and promote widespread data dissemination among research communities cutting across clinical, translational and epidemiological domains, primarily through collaboration with the established CardioVascular Research Grid (CVRG).