Convergence Accelerator Phase I (RAISE): A Multi-Scale Open Knowledge Network for Precision Medicine

NSF融合加速器支持基于团队的多学科努力，以应对国家重要性的挑战，并在不久的将来显示出交付成果的潜力。这个项目的目标是将不同领域的交叉信息联系起来，并以一种导致创新和新知识的方式将其整合起来。例如，如果美国财政部能够从所有国内外银行、金融市场甚至社交网络获得私人金融信息，他们或许就能预见到2008年的大衰退。该项目使用相同的原理，但专注于计算工具和生物医学信息，有可能推动一系列生物医学学科的发现。具体地说，这个融合加速器第一阶段项目将通过整合来自多个生物医学数据库的信息来创建生物医学知识引擎。随着人工智能的使用，它将使研究人员拥有加快基础生物医学研究和药物发现的工具，以及医生具有潜在的前所未有的患者洞察力。该计划的更广泛的影响和潜在的社会利益取决于其促进和民主化获取高度专业化的、但公开的生物医学重要信息的能力。目前，大多数生物医学数据源都被隔离在专门的门户中，集成最少。这一融合项目整合了广泛的专业知识，从基因组学、药理学和病人护理，到计算机科学、深度数据科学和认识论。该小组由来自学术机构(加州大学旧金山分校)、政府(劳伦斯·利弗莫尔国家实验室)、非营利组织(系统生物学研究所)和商业实体(谷歌)的研究小组组成，在项目期间将与其他组织接触。虽然这些团队一直在较小范围的项目中合作(主要是以结对的方式)，但该项目打算围绕一个更大的、共同的目标来明确团队。第一阶段工作中规定的可交付成果将有助于建立一个由数十亿个概念组成的知识网络(图)，这些概念通过具有生物意义的关系联系在一起。该图表将向公众开放，可通过手动或自动搜索访问，其内容可以更新或修改，就像今天的万维网一样。将来自多个领域的海量信息整合到一个广泛的知识网络中，首次开启了在通常不相互作用的学科(如内科医学和分子生物学)之间通过计算导航图表的可能性。此外，由于将要开发的知识图谱包含特定领域的知识，但不包含单个患者数据，因此不会引起隐私问题。这个项目的智力价值在于将团队成员之间的领域专业知识进行了重要的集成，并利用了由其他公共和私人资助的工作创建的数据资源。该项目为一个资源奠定了基础，该资源可能会让研究人员和从业者以谷歌搜索的方式立即获得所有相关的“生物医学事实”的整体，有可能改变生物医学研究、药物开发，甚至全国各地的医学实践方式。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The NSF Convergence Accelerator supports team-based, multidisciplinary efforts that address challenges of national importance and show potential for deliverables in the near future. The goal of this projects is to connect intersecting information across distinct domains and integrate it in a way that leads to innovation and new knowledge. For instance, if the US Treasury had access to private financial information from all domestic and foreign banks, financial markets, and even social networks, they might have been able to foresee the Great Recession of 2008. This project uses the same principle but is focused on computational tools and biomedical information with the potential to advance discovery across a range of biomedical disciplines. Specifically, this Convergence Accelerator Phase I project will create a biomedical knowledge engine by integrating information from multiple biomedical databases. With the use of artificial intelligence, it will empower researchers with the tools to accelerate basic biomedical research as well as drug discover, and doctors with potentially unprecedented patient insights. The broader impact and potential societal benefit of this program lies on its ability to facilitate and democratize access to highly-specialized, yet publicly available information of biomedical importance. Currently, the majority of biomedical data sources are secluded within dedicated portals with minimal, if any, integration. This convergent project integrates wide-ranging expertise, from genomics, pharmacology and patient care, to computer science, deep data science, and epistemology. This team comprises research groups from academic institutions (University of California San Francisco), government (Lawrence Livermore National Labs), non-profit (Institute for Systems Biology) and commercial entities (Google), and during the project will engage with other organizations. While these groups have been collaborating (mostly in a pair-wise fashion) in projects with a smaller scope, this project intends to crystallize the team around a larger, common objective. The deliverables specified in this Phase I effort will contribute to the creation of a knowledge network (graph) composed of billions of concepts connected by biologically meaningful relationships. The graph will be open to the general public, accessible via manual or automated searches, whose content can be updated or modified, much like the world wide web is today. Integrating vast amounts of information from multiple domains to an extensive knowledge network opens for the first time the possibility to computationally navigate the graph across disciplines that normally do not interact (like internal medicine and molecular biology). Furthermore, because the knowledge graph that will be developed contains domain-specific knowledge but not individual patient data, it does not pose privacy concerns. The intellectual merit of this project lies in the significant integration of domain expertise across team members and the leveraging of data resources that have been created by other publicly and privately funded efforts. The project lays the groundwork for a resource that will potentially afford researchers and practitioners immediate access to the totality of all relevant "biomedical facts" - in the style of Google search, with the potential to change biomedical research, drug development and even the way medicine is practiced across the nation.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.