Latent variable analysis in cognitive aging executive

认知老化执行者的潜变量分析

• 批准号: 7142300
• 负责人: ROCHELLE E TRACTENBERG
• 金额: $ 12.61万
• 依托单位: GEORGETOWN UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-09-01 至 2011-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7142300
• 关键词: Alzheimer' s disease; aging; behavioral /social science research tag; brain disorder diagnosis; clinical research; cognition disorders; data collection methodology /evaluation; data quality /integrity; dementia; developmental psychology; executive function; functional ability; human data; human old age (65+); human subject; mathematical model; model design /development; neuropsychological tests; psychological aspect of aging; psychometrics; short term memory

DESCRIPTION (provided by applicant): The symptom domains of greatest interest in Alzheimer's disease (AD), cognition, function, and behavior, are also associated with executive function (EF). But EF cannot be directly observed; these symptom domains and EF have unknown measurement error in their assessment, and unknown interrelationships. Precise assessment of each area is important for efficient clinical studies of AD interventions and for public health purposes like screening for AD and dementia. Unlike regression analysis, latent variable analysis (LVA) methods can increase precision by explicitly modeling measurement error, and can model constructs (like EF) that are not directly observable. Additional precision will be derived from determining whether EF contributes significant explanatory power to models of cognitive aging and AD. This K01 therefore has two specific aims. Aim 1: Model measurement error in instruments that are common in clinical trials for AD using a Bayesian Network to differentiate measurement error from true change for non-demented elderly, persons with AD, and persons with incipient AD. Aim 2: Construct a statistical model of EF for persons with AD and cognitively normal elderly controls in order to: a. evaluate the latent factor structure in EF; and b. compare fits of latent variable models of EF. The best fitting model will address the hypothesis that EF contributes significant explanatory power for variance in cognitive, functional and behavioral symptoms and will suggest whether this is true for both normal elderly and persons with AD. These aims will contribute to precision and modeling of assessment in normal aging and AD; and to a better understanding of EF as an outcome in aging research and clinical studies of AD. The proposed K01 will develop deep and broad sets of skills in assessment and analysis, to be achieved with training in design and conduct of clinical trials, measurement theory, neuropsychology, and advanced statistical/LVA modeling. Mentoring from an experienced clinical trialist, Paul Aisen, MD, will be supplemented with expert consultation in these domains throughout the five year grant. The PI is a junior investigator in who wishes to extend her expertise in statistics to include LVA methods, and her background in cognitive science to include neuropsychological assessment and measurement. This K01 award will achieve these and so will move the PI closer to her long-term goal of becoming an independent researcher in cognitive aging and AD.

描述（由申请人提供）：阿尔茨海默病（AD）、认知、功能和行为中最令人感兴趣的症状领域也与执行功能（EF）相关。但EF不能直接观察；这些症状域和 EF 在评估中存在未知的测量误差以及未知的相互关系。 每个领域的精确评估对于 AD 干预措施的有效临床研究以及 AD 和痴呆症筛查等公共卫生目的非常重要。与回归分析不同，潜变量分析 (LVA) 方法可以通过显式建模测量误差来提高精度，并且可以对不可直接观察的构造（如 EF）进行建模。通过确定 EF 是否对认知衰老和 AD 模型具有显着的解释力，将获得额外的精度。因此，这款 K01 有两个具体目标。目标 1：使用贝叶斯网络对 AD 临床试验中常见的仪器测量误差进行建模，以区分测量误差和非痴呆老年人的真实变化 AD 患者和早期 AD 患者。目标 2：为 AD 患者和认知正常的老年人对照构建 EF 统计模型，以便：评估 EF 中的潜在因子结构；和b。比较 EF 潜变量模型的拟合度。最佳拟合模型将解决这样的假设：EF 对认知、功能和行为症状的差异具有显着的解释力，并将表明这对于正常老年人和 AD 患者是否都成立。这些目标将有助于提高正常衰老和 AD 评估的精确度和建模；更好地理解 EF 作为衰老研究和 AD 临床研究的结果。拟议的 K01 将培养深入而广泛的评估和分析技能，通过临床试验设计和实施、测量理论、神经心理学和高级统计/LVA 建模方面的培训来实现。在五年资助期间，经验丰富的临床试验专家 Paul Aisen 医学博士的指导将得到这些领域专家咨询的补充。 PI 是一名初级研究员，她希望扩展她在统计学方面的专业知识，包括 LVA 方法，以及她在认知科学方面的背景，包括神经心理学评估和测量。 K01 奖项将实现这些目标，从而使 PI 更接近成为认知衰老和 AD 领域独立研究员的长期目标。