Administrative

我们将建立从大数据中发现生物医学知识的因果建模和发现中心（CCMD），作为匹兹堡大学（皮特;牵头机构），卡内基梅隆大学（CMU）和耶鲁大学（耶鲁大学）之间的合作。CCMD将开发、验证和传播基于因果贝叶斯网络的方法、工具和软件，这将使更广泛的科学界能够有效地查询大型成像、基因组和临床（表型）数据，并获得有关观察到的现象的因果关系的知识。我们选择了三个驱动生物医学问题（DBP）的癌症途径，肺部疾病和脑功能磁共振成像作为算法和软件开发的平台。我们的总体目标是开发和实现广泛使用，一套完整的互操作软件和应用程序编程接口，这将有助于推进BD 2K研究和教育。CCMD的活动将分为三个主要部分。数据科学研究（DSR），培训和联盟。CCMD将由3名PI共同领导，与中心执行委员会协商，每位PI将共同领导一个活动部分。行政核心（AdminCore）将监督CCMD的整体运作，以确保参与机构的团队之间进行富有成效的互动，有效整合DSR和CCMD的培训活动，有效开发有利于更广泛研究界的CMD计算工具，培训下一代数据科学家，掌握解决各种BD 2K问题所需的CMD技能。管理核心还将继续监测和评估CCMD三个组成部分的进展情况，制定战略计划，以提高研究、培训和合作活动的效用和质量。我们将使用详细的逻辑模型和相关指标作为其评估框架，并将与中心内部和外部顾问委员会进行磋商。管理核心将负责确保获得足够的基础设施和资源，以有效开发和传播CMD工具，使数据科学和生物医学研究界受益，特别是其他BD 2K中心。

We will establish our Center for Causal Modeling and Discovery (CCMD) of Biomedical Knowledge from Big Data as a collaboration among the University of Pittsburgh (Pitt; lead institution), Carnegie Mellon University (CMU), and Yale University (Yale). The CCMD will develop, validate, and disseminate methods, tools, and software based on Causal Bayesian Networks, which will enable the broader scientific community to effectively interrogate large imaging, genomic, and clinical (phenotype) data and derive knowledge on the causality of observed phenomena. We selected three Driving Biomedical Problems (DBPs) on cancer pathways, lung diseases, and brain fMRI as a platform for algorithm and software development. Our overarching goal is to develop and enable the broad usage of, a complete suite of interoperable software' and application programming interface that will help advance BD2K research and education. The activities of the CCMD will be organized in three major components. Data Science Research (DSR), Training and Consortium. The CCMD will be co-led by 3 PIs, each of whom will co-lead an activity component, in consultation with the Center Executive Committee. The Administrative Core (AdminCore) will oversee the overall operation of the CCMD to ensure productive interaction between the teams at the participating institutions, effective integration of the DSR and training activities of the CCMD, efficient development of a computational tools for CMD that will benefit the broader research community, and training of next generation of data scientists equipped with CMD skills necessary to tackle a variety of BD2K problems. The AdminCore will also continually monitor and evaluate the progress made in the three CCMD components, making strategic plans to increase the utility and quality of research, training, and collaborative activities. We will use a detailed logic model and associated metrics as its evaluation framework and will consult with the Center Internal and External Advisory Boards. The AdminCore will be responsible for ensuring access to adequate infrastructure and resources for efficient development and dissemination of CMD tools to benefit the data science and biomedical research communities, in particular the other BD2K Centers.