Data Science & Management Core

摘要 数据科学与管理核心(Timothy Verstinen，CL) 该项目旨在阐明行为和社会环境影响的神经生物学。 中年成年人患心血管疾病的风险。满足项目的特殊分析需求 该项目需要将传统统计方法与尖端机器学习和数据相结合 大型高维数据集的科学工具(即多变量神经成像、生物学和行为学 在多个时间点收集的数据)。因此，核心C是为这些新的分析挑战量身定做的。 具体地说，核心C的综合方法包括组合方法，这些方法可以(I)识别特定的 通过对大脑进行数十万次测量得出的功能性或结构性大脑网络 可靠地预测生物、行为和心血管健康相关变量，以及(Ii)生成适当的 来自这些大脑测量的度量可以集成到使用神经的结构方程模型中， 预测心血管疾病风险的介体和标记物随时间变化的行为和生理因素。至 为此，Core C将聘用：(I)机器学习、多层次建模方法和生物统计学方面的专家 用于生成脑表型指标(目标1)和开发心血管疾病风险出现的预期模型 (目标2)；和(2)有经验的数据管理员制定安全收集和维护 项目数据(目标4)。核心C还将与管理核心(核心A)合作，以确保有效地流动 通过定期更新测量数据的接收情况，通过方案项目获取数据 核心(核心B)和3个项目(P)，以及向每个P分发数据。 Timothy Verstinen(一位受过博士培训的认知神经学家，在量化方面具有高级专业知识 建模和机器学习)，并得到Aarti Singh博士(统计学家)的协调努力 和机器学习方面的权威)，核心C将(I)提供集中研究支持，(Ii)提供高级 统计专门知识和统一数据分析；(3)就新的统计方法进行协商；(4)提供培训 关于统计方法；和(V)存档和公开共享数据。国际上的外部咨询公司 公认的统计学家和量化方法学家也将为特定的 预期的分析挑战类型(例如，纵向结构方程建模、缺失数据、 纵向数据的多层次建模、心理测量学和潜变量建模)。在其面板上绘制 专家统计和方法顾问和数据科学团队，Core C也将提供教育 为项目研究人员、受训人员和科学工作者提供创新分析和研究方法的服务 社区(目标3)。最后，配合A芯，C芯将实施P01数据共享计划，以 使所有项目数据在公共领域中可用。核心成员将定期会面并进行协作 与P个调查员合作，传播其协同活动所产生的研究成果。

ABSTRACT Data Science & Management Core (Timothy Verstynen, CL) This Program Project aims to explicate the neurobiology of behavioral and socio-environmental influences on risk for cardiovascular disease (CVD) in midlife adults. Meeting the special analytic demands of the Program Project requires integrating traditional statistical approaches with cutting-edge machine learning and data science tools for large, high-dimensional data sets (i.e., multivariate neuroimaging, biological, and behavioral data collected at multiple time points). Accordingly, Core C is tailored to these novel analytical challenges. Specifically, the integrative approach of Core C consists of combining approaches that can (i) identify specific functional or structural brain networks from hundreds of thousands of measurements across the brain that reliably predict biological, behavioral, and cardiovascular health-related variables and (ii) generate appropriate metrics from these brain measurements that can be integrated into structural equation models that use neural, behavioral, and physiological factors to predict changes in mediators and markers of CVD risk over time. To this end, Core C will employ: (i) experts in machine learning, multi-level modeling methods, and biostatistics for generating brain phenotype metrics (Aim 1) and developing prospective models of CVD risk emergence (Aim 2); and (ii) experienced data managers to develop protocols for the secure collection and maintenance of Project data (Aim 4). Core C will also work with the Administrative Core (Core A) to ensure efficient flow of data through the Program Project by providing regular updates on the receipt of data from the Measurement Core (Core B) and the 3 Projects (Ps), as well as data distribution to each P. Under the leadership of Dr. Timothy Verstynen (a doctorally-trained cognitive neuroscientist with advanced expertise in quantitative modeling and machine learning) and as reinforced by the coordinating efforts of Dr. Aarti Singh (a statistician and authority in machine learning), Core C will (i) provide centralized research support, (ii) provide advanced statistical expertise and harmonize data analyses; (iii) consult on novel statistical methods; (iv) deliver training on statistical methods; and (v) archive and publically share data. An external consultancy of internationally recognized statisticians and quantitative methodologists will also provide ongoing consultation for the specific types of analytic challenges anticipated (i.e., longitudinal structural equation modeling, missing data, multilevel modeling of longitudinal data, psychometrics and latent variable modeling). Drawing on its panel of expert statistical and methodological consultants and the data science team, Core C will also offer educational services on innovative analytic and research methods to Program investigators, trainees, and the scientific community (Aim 3). Lastly, in coordination with Core A, Core C will implement the P01 data sharing plan to make all Project data available in the public domain. Core members will meet regularly and will collaborate with P investigators on the dissemination of research findings emanating from their synergistic activities.