Metadata applications on informed content to facilitate biorepository data regulation and sharing

摘要 生物储存库对于实现现代分子基础研究至关重要，这将推动生物技术的发展。 新一代靶向诊断和治疗以及个性化药物，以改善临床 患者的结果。研究期间收集的数据和生物标本资源的使用受到限制 研究参与者向研究团队提供的知情同意书、研究方案文件，以及 IRB本身对研究施加的限制。目前缺乏一个共同的模式， 同意，限制如何容易数据（和研究标本）从多个研究项目，或多个 机构可以结合起来进行大规模的回顾性研究。手动检查和协调 来自不同生物库的数百万份知情同意书可能变得昂贵， 不可调和的问题。应用适当的元数据来支持一套复杂的监管、法律的 隐私和安全要求的过程和信息流所涉及的监管研究是一个领域， 发展的早期阶段。过程中涉及的复杂的法律的和技术要求 挑战我们有效建立信息系统的能力，以支持研究数据、标本 和其他大规模的研究文物。此外，研究中涉及的许多监管过程是 仍然基于纸质工作流程。我们通过开发合适的基于机器的元数据来验证这一点 代表监管过程，重点是知情同意和相关文件， 加强监管机构（如机构审查委员会）的能力，以1）以更广泛的方式审查提案 简化的方式，2）有潜力提供一个更全面的了解风险沿着 多轴，以及3）为数据共享和信息发布策略提供正式的和可计算的基础。 更具体地说，我们将专注于三个具体目标：1）开发符合标准的元数据 知情同意的表示; 2）开发基于NLP的知情同意自动注释工具 文档;以及3）评估用于语义地表示域的基于元数据本体的方法 并展示其回答能力问题的能力。我们提出的方法是新颖的， 以下原因：1）据我们所知，它提供了第一个元数据本体，以表示 知情同意空间，同时考虑美国的共同规则; 2）它结合了自然语言处理 技术与基于本体的方法进行语义注释以及本体丰富;和3）它 让利益相关者参与本体开发和评估过程，并使用能力问题， 验证本体的覆盖率。

ABSTRACT Biorepositories are critical to enabling modern molecular-based research that will drive the development of a new generation of targeted diagnostics and therapies as well as personalized medicine to improve clinical outcomes for patients. The use of data and biospecimen resources collected during research are constrained by the informed consent that research participants give to research teams, research protocol documents, and the constraints imposed on the research by the IRB itself. Currently there is a lack of a common model of consent that limits how easily data (and research specimens) from multiple research projects, or multiple institutions can be combined for large-scale retrospective studies. Manually examining and reconciling potentially millions of informed consent forms from different biobanks becomes an expensive and possibly irreconcilable problem. The application of suitable metadata in support of the complex set of regulatory, legal, privacy and security requirement processes and information flows involved in regulated research is a field in an early phases of development. The complicated legal and technical requirements involved in the processes challenge our ability to effectively build information systems that support sharing of research data, specimens and other research artifacts at scale. Additionally, much of the regulatory processes involved in research are still based on paper-based workflows. We posit that by developing suitable machine-based metadata representation of regulatory processes focusing on informed consents and the associated documents would enhance the ability of regulatory bodies such as Institutional Review Boards to 1) review proposals in a more streamlined fashion, 2) have the potential to provide a more comprehensive understanding of risk along multiple axes, and 3) provide a formal and computable basis for data sharing and information release policies. More specifically we will focus on three specific aims: 1) Develop standard-conforming metadata representations of informed consent; 2) Develop NLP-based automatic annotation tool for informed consent documents; and 3) Evaluate the metadata-ontology-based approach for semantically representing the domain and demonstrate its capacity for answering competency questions. Our proposed approach is novel for the following reasons: 1) it provides the first metadata ontology, to the best of our knowledge, to represent the informed consent space while considering the US common rules; 2) it combines natural language processing technologies with ontology-based approaches for semantic annotation as well as ontology enrichment; and 3) it engages stakeholders in the ontology development and evaluation process and uses competency questions to verify the coverage of the ontology.