Data Management Core

项目概要 西北 HeartShare 数据翻译中心 (DTC) 数据管理核心将利用先前的 在机器学习领域识别新型心力衰竭方面的领导力和经验 保留射血分数 (HFpEF) 亚型，以及多方面领先数据管理的跟踪记录 中心研究，以促进多维度、多组学分析（包括机器学习/人工 智能）用于整个 HeartShare 计划。 HeartShare 数据管理核心包括一个高度 由来自不同学科的个人组成的综合团队，具有数据协调中心运营方面的专业知识， 统计学、机器学习、组学和 HFpEF。这些专家将共同努力实现以下目标 具体目标：(1) 制定数据汇集、清理和统一所有数据的程序和标准 来自队列和试验；将数据汇编为研究界的共享资源；并托管这些数据 BioData Catalyst 作为 HeartShare 回顾性部分的一部分； (2) 整合自我报告数据， 来自移动健康 (mHealth) 技术的家庭监控数据、环境数据、电子健康记录 (EHR) 数据（包括自然语言处理 [NLP] 技术）、深度表型数据和结果 将数据转化为 HeartShare 未来组件的核心数据资源； (3)建立综合生物样本库 和成像存储库资源；与 TOPMed 协调这些资源；并促进自动化深度 学习与 HeartShare 相关的图像分析； (4) 规划和促进高级分析（多组学 和机器学习分析）来识别 HFpEF 亚型以及这些亚型背后的病理生物学基础 亚型。数据管理核心还将确保高效的可视化、报告和通信 调查结果；我们将与我们的数据门户核心中的 BioData Catalyst 专家合作，以公开提供 将数据存储到云存储中，在 BioData Catalyst 中实施治理，并使工具和工作流程云端化 兼容。

PROJECT SUMMARY The Northwestern HeartShare Data Translation Center (DTC) Data Management Core will leverage prior leadership and experience in the field of machine learning for the identification of novel heart failure with preserved ejection fraction (HFpEF) subtypes, along with a track record of leading data management for multi- center studies, to facilitate multi-dimensional, multi-omics analyses (including machine learning/artificial intelligence) for the entire HeartShare program. The HeartShare Data Management Core includes a highly integrated team of individuals from a variety of disciplines with expertise in data coordinating center operations, statistics, machine learning, omics, and HFpEF. These experts will work in concert to achieve the following specific aims: (1) to develop procedures and standards for data pooling, cleaning, and harmonization of all data from cohorts and trials; compile the data as a shared resource for the research community; and host these data on BioData Catalyst as part of the retrospective component of HeartShare; (2) to integrate self-reported data, home monitoring data from mobile health (mHealth) technologies, environmental data, electronic health record (EHR) data (including natural language processing [NLP] techniques), deep phenotyping data, and outcome data into a core data resource for the prospective component of HeartShare; (3) to create integrated biorepository and imaging repository resources; coordinate these resources with TOPMed; and facilitate automated deep learning analyses of images relevant to HeartShare; and (4) plan and facilitate advanced analytics (multi-omics and machine learning analyses) to identify HFpEF subtypes and the pathobiological basis underlying these subtypes. The Data Management Core will also ensure efficient visualization, reporting, and communication of findings; and we will work with BioData Catalyst experts in our Data Portal Core to transition publicly available data into cloud storage, implement governance within BioData Catalyst, and make tools and workflows cloud compatible.