GHUCCTS N3C COVID data mapping

摘要 充分利用现有数据和资源的一个主要挑战是卫生保健的复杂性质 数据，以及数据来源的异质性(包括非结构化的临床记录)，以及缺乏 标准。缺乏标准排除了跨平台和机构之间的语义互操作性。 相反，当前的方法利用资源密集型自然语言过程来提取、转换和 将来自不同来源的数据关联起来进行分析。提高翻译科学水平，加快研究 为了改善患者结果，许多新的和创新的研究正在利用大量可用的数据 通过标准化和共享数据倡议。随着当前计算和健康数据分析的进步 工具、方法和访问，并使数据更有意义、更开放和更容易访问，研究研究 超越了传统的追溯性报告，转向了务实的干预和预测能力。正在进行中 努力的重点是利用共同的数据标准和模型，如观察性医疗结果 合作伙伴(OMOP)标准-由观察卫生数据科学和信息学(OHDSI)DefiNed 财团，并被美国国立卫生研究院和PCORI接受为典范-将带头发现 文本叙事，强制数据标准化，促进可伸缩性和共享。OHDSI公共数据 模型(CDM)使数据更有意义、更开放、更易于访问，从而推动了翻译科学和 允许跨不同的数据源一致地开发预测模型。国家COVID 队列协作(N3C)、ACT、BD2K-NIH数据共享、国家健康数据中心(CD2H)、 以及其他一些努力将导致新的发现和明智的决策，由 数据科学，并以成熟的大数据技术为支撑。我们建议设计和建立小说， 可扩展、标准化的大数据流程，可大规模提取原始电子病历 用于观察性研究的数据集。该项目将开发一个安全的基于云的环境来托管这些 数据，以及支持观测研究的应用程序编程和图形用户界面 利用这些资源的研究。通过这些手段，我们将降低数据标准化的障碍， 可重复分析的注释和共享，并开始强制执行完整的语义和句法 数据生态系统中资源之间的互操作性。这一努力将使我们的调查人员能够研究 医疗干预的效果和预测患者的健康结果并产生经验证据 建立观测分析最佳做法所需的基础。

Abstract A major challenge to full utilization the available data and resources has been the complex nature of health data, and heterogeneity of data sources (including unstructured clinical notes) combined with a lack of standards. The lack of standards precludes semantic interoperability across platforms and between institutions. Instead, current approaches utilize resource intensive natural language processes to extract, transform, and correlate data from different sources for analysis. To improve translational science and accelerate research to improve patient outcomes, many new and innovative studies are leveraging large volumes of available data through standardized and shared data initiatives. With current advances in computing and health data analysis tools, methods and access, and to make data more meaningful, open, and accessible, research studies have moved beyond traditional retroactive reporting to pragmatic interventions and predictive capabilities. Ongoing efforts focus on exploiting common data standards and models such as the Observational Medical Outcomes Partnership (OMOP) standard—defined by the Observational Health Data Sciences and Informatics (OHDSI) consortium, and accepted as canon by both the NIH and PCORI— will lead the way to discover insights in textual narrative, enforce data standardization, and promote scalability and sharing. The OHDSI Common Data Models (CDM) makes data more meaningful, open, and accessible, which drives translational science and allows for consistent development of predictive models across different data sources. The National COVID Cohort Collaborative (N3C), ACT, BD2K-NIH Data Commons, the National Center for Data to Health (CD2H), and others are among the efforts that will lead to new discoveries and informed decision making, driven by data science and undergirded by mature Big Data technologies. We propose to design and establish novel, scalable, and standardized big data processes to massively abstract the raw electronic medical record datasets for observational studies. This project will develop a secure cloud-based environment to host these data, as well as the application programming and graphical user interfaces to support observational research studies leveraging these resources. By these means we will reduce the barriers to data standardization, annotation and sharing for reproducible analytics and begin to enforce complete semantic and syntactic interoperability between the resources in the data ecosystem. This effort will enable our investigators to study the effects of medical interventions and predict patients' health outcomes and generate the empirical evidence base necessary to establish best practices in observational analysis.