NeuroMAP Phase II - Data Management and Statistics Core

项目摘要：数据管理和统计核心 研究项目负责人 (RPL) 的一项为期 3 年的项目，旨在与 精神科目标人群并获得 R 级补助金的试点数据在设计上可能具有挑战性， 数据管理和处理，以及统计分析视角。数据管理与统计 (DMS) Core 将确保最严格的研究设计、社区标准数据的实施 管理和处理协议，以及尖端数据科学算法的应用 最大限度地提高样本外预测性能和评估作用机制的能力。核心将 与 RPL 和试点项目研究人员合作，促进识别和验证疾病缓解措施 与情绪和焦虑症相关的过程（DMP）。这将大大提高该系统的实用性 RPL 进行的研究，用于制定目标并在此基础上提出假设 初步数据并用于设计未来的研究，从而使 R01 级应用更有可能成功 以及更具竞争力和可资助性。该核心提供的服务包括： (1) 咨询 与专家数据科学家合作，他们将与研究人员合作开发和实例化操作环境 优化数据使用和分析； (2) 该核心开发的程序和程序，以 满足用户的刺激呈现、数据管理和统计需求。数据管理 组件将有助于保证可靠、高效地获取和处理数据 使用我们的可扩展数据管理基础设施。服务将从研究设置开始，包括 行为范例的实现和配置，将原始数据转换为标准的管道（例如， 脑成像数据结构：BIDS）格式，根据需要定期审核和共享。该核心将提供 标准管道来提取通用数据元素和质量指标，并促进访问和使用 该研究所的计算基础设施。核心的统计部分将侧重于开展研究 适用于评估 DMP 的无偏效应和预测性能的设计和分析程序 （例如，威胁敏感性、厌恶性内感受期间的回避、重复的消极思维）对心理健康的影响 结果。由于这些 DMP 将在多个分析层面进行检查（症状、行为、生理学、 电路和分子），研究设计和分析需要整合复杂的多方法关联 并且需要考虑关联中潜在的偏差，例如，由于选择、测量误差和/或 令人困惑。该核心将重点关注多级模型、因果推理和机器学习预测 考虑变异来源（例如嵌套数据）和混杂因素（例如混杂偏差），同时提供 最大解释性和样本外预测性能。该核心的产品将发挥重要作用 利用 NeuroMAP 项目产生的数据开发后续 R 级研究。

PROJECT SUMMARY: Data Management and Statistics Core A 3-year project for a Research Project Leader (RPL) to conduct experimental human subjects research with psychiatric target populations and to obtain pilot data for an R-level grant can be challenging from a design, data management and processing, and statistical analysis perspective. The Data Management and Statistics (DMS) Core will ensure the highest rigor of study design, the implementation of community-standard data management and processing protocols, and the application of cutting-edge data science algorithms that maximize out-of-sample prediction performance and power for assessing mechanisms of action. The Core will work with RPLs and pilot project investigators to facilitate identifying and validating disease-modifying processes (DMPs) that are relevant for mood and anxiety disorders. This will greatly enhance the utility of the research produced by RPLs for use in formulating aims and developing hypotheses based on these preliminary data and for designing future studies, thereby making R01-level applications more likely to succeed as well as being more competitive and fundable. Services provided by this Core consist of: (1) consultations with expert data scientists who will work with investigators to develop and instantiate an operating environment that optimizes data use and analytics; and (2) procedures and programs developed by this Core to accommodate users' stimulus presentation, data management and statistical needs. The data management component will be instrumental in guaranteeing that data are acquired and processed reliably and efficiently using our scalable data management infrastructure. Services will begin at study setup and include implementation and configuration of behavioral paradigms, pipelines to convert raw data into standard (e.g., Brain Imaging Data Structure: BIDS) format, periodic auditing and sharing as needed. This Core will provide standard pipelines to extract common data elements and quality metrics and to facilitate access and usage of the institute's computing infrastructure. The statistics component of the Core will focus on developing study designs and analytic procedures applicable to assessing unbiased effects and predictive performance of DMPs (e.g., threat sensitivity, avoidance during aversive interoception, repetitive negative thinking) on mental health outcomes. As these DMPs will be examined on several levels of analysis (symptoms, behavior, physiology, circuits, and molecules), study designs and analyses will need to integrate complex multi-method associations and will need to account for potential biases in associations, e.g., due to selection, measurement error, and/or confounding. This Core will focus on multilevel models, causal inference and machine learning prediction that account for sources of variation (e.g., nested data) and confounding (e.g., confounding bias) while providing maximal explanatory and out-of-sample prediction performance. The products of this Core will be instrumental in developing follow-up R-level research leveraging data produced by NeuroMAP projects.