Statistical Methods for Developing RDoC-based Multidimensional Profiles

描述(申请人提供)：梅雷迪思·洛兹·华莱士博士的长期研究目标是成为一名精神病学生物统计学家，开发和应用新方法来帮助研究人员实现NIMH研究领域标准(RDoC)项目的目标。RDoC项目表明，精神障碍的理想特征可能是多维的体征和症状(例如，大脑电路、生理、行为和自我报告)；然而，这种新的范式构成了巨大的挑战，因为大多数现有的统计方法没有设计来适应由此产生的数据类型的混合。华莱士博士提议的导师研究科学家发展奖(K01)是朝着她的长期研究目标迈出的第一步；它将为她提供开发统计方法所需的培训和支持，使RDoC项目成为更可行的研究框架。华莱士博士K01的总体目标是开发和应用结合多种类型密集测量数据的新的聚类方法。在目标1中，华莱士博士将开发集群方法，将通过多导睡眠图、活动记录仪和每日自我报告测量捕获的密集测量数据合并在一起。这些方法将在两个不同的理论框架下开发，以增强它们在K01中具体阐述的那些应用之外的普适性。在目标2中，华莱士博士将把她的方法应用到一个数据库中，该数据库包含1000多个个体的多个维度的观察结果，这些个体存在与失落、焦虑和睡眠/清醒调节相关的不同程度的干扰。这些方法的应用将揭示可能跨越当前基于DSM的诊断的多维轮廓。对这些特征的研究将阐明体征和症状的多个维度在多大程度上处于一个真正的连续体或分为不同的类别。尽管这位候选人在生物统计学方面的背景已经为她开始这项研究做好了准备，但这项研究的成功最终取决于三个领域的额外培训：1)聚类和密集测量数据方法的数学基础，2)负价系统(例如，丢失和焦虑)，以及3)睡眠/清醒调节。这种培训将通过教程、定向阅读、正式课程作业、动手临床工作和专业会议来实现。为了实现她的培训目标，华莱士博士组建了一支由临床和方法学导师和顾问组成的强大团队。临床指导团队由埃伦·弗兰克博士领导，他是国际公认的情绪相关障碍专家。方法论指导团队由萨蒂什·艾扬格博士领导，他是精神病学统计领域的领导者，擅长对密集测量的数据进行分类和建模。这个项目带来的培训和研究将为华莱士博士提供在K01的第四年提交R01所需的经验和技能。R01将允许华莱士博士扩展她提出的方法，并将其应用于在其他维度和研究领域的基础上开发基于RDoC的多维概况。

DESCRIPTION (provided by applicant): Dr. Meredith Lotz Wallace's long-term research goal is to become a psychiatric biostatistician who develops and applies novel methods to aid researchers in attaining the goals of the NIMH Research Domain Criteria (RDoC) project. The RDoC project suggests that mental disorders may be ideally characterized by multiple dimensions of signs and symptoms (e.g., brain circuitry, physiology, behavior, and self-report); however, this new paradigm poses substantial challenges because most existing statistical methods were not designed to accommodate the resulting mixture of data types. Dr. Wallace's proposed Mentored Research Scientist Development (K01) award is the first step towards her long-term research goal; it will provide the training and support she needs to develop statistical methods that will make the RDoC project a more viable research framework. The overall aim of Dr. Wallace's K01 is to develop and apply novel clustering methods that incorporate multiple types of intensively measured data. In Aim 1, Dr. Wallace will develop clustering methods that incorporate intensively measured data captured through polysomnography, actigraphy, and daily self-report measures. These methods will be developed under two different theoretical frameworks to enhance their generalizability to applications beyond those addressed specifically in the K01. In Aim 2, Dr. Wallace will apply her methods to a data base containing multiple dimensions of observations from over 1000 individuals with a variety of levels of disturbances related to loss, anxiety, and sleep/wake regulation. The methods' application will reveal multidimensional profiles that may cut across the current DSM-based diagnoses. A study of these profiles will clarify the extent to which the multiple dimensions of signs and symptoms lie on a true continuum or separate into discrete classes. Although the candidate's background in biostatistics has primed her to begin this research, its successful completion ultimately hinges on additional training in three areas: 1) mathematical foundations of methods for clustering and intensively measured data, 2) negative valence systems (e.g., loss and anxiety), and 3) sleep/wake regulation. This training will be achieved through tutorials, directed readings, formal coursework, hands-on clinical work, and professional conferences. To achieve her training goals, Dr. Wallace has assembled a strong team of clinical and methodological mentors and consultants. The clinical mentoring team is led by Dr. Ellen Frank, an internationally recognized expert in mood-related disorders. The methodological mentoring team is led by Dr. Satish Iyengar, a leader the field of psychiatric statistics with expertise in clustering and modeling intensively measured data. The training and research resulting from this project will provide Dr. Wallace with the experience and skills she needs to submit an R01 in the fourth year of the K01. This R01 will allow Dr. Wallace to expand on her proposed methods and apply them to develop RDoC-based multidimensional profiles based on additional dimensions and domains of study.