|
Abstract:
|
Recent research has identified cognitive decline as a precursor to mild cognitive impairment (MCI) and Alzheimer's disease (AD). In this study, we explore various cognitive endpoints measured in elderly subjects in the Rush Memory and Aging Project database. The objective is to better understand different cognitive tests and their association with the onset of MCI and/or AD, and potentially help design AD prevention studies with respect to endpoints selection. We propose to use item response theory (IRT) modeling approach, which enables one to estimate both test-level parameters and patient-level cognitive abilities. The estimated difficulty, discrimination, and Fisher information of each test are compared. Separate IRT models are fitted to progressors and non-progressors, and patients are categorized based on the likelihood from two models. A longitudinal IRT model is explored to investigate which cluster of tests detects an early decline in cognitive ability for progressors. Our study provides a better understanding of various cognitive tests and helps trial designer to make an informed decision when it comes to what endpoints to use in AD prevention studies.
|
