PROJECT 2: SUMMARY/ABSTRACT Project 2 is entitled “Novel Measurement Approaches to Preclinical AD”, and is designed to capitalize on the fact that the subjects in the BIOCARD study have been followed for up to 27 years. This offers an unparalleled opportunity to identify biomarkers, together with their combinations, that can predict which cognitively normal individuals will progress to mild cognitive impairment, as well as the changepoints in these biomarkers during preclinical Alzheimer’s disease. However, this wealth of data also presents many analytical challenges. These challenges include: how to optimally or efficiently combine biomarker measurements and how to estimate their changepoints, together with rates of change in the biomarkers, in relation to the incidence of mild cognitive impairment. The focus of this project is therefore to examine these essential biostatistical issues for the study of preclinical AD, using data from the BIOCARD study. The specific aims include: (1) To develop machine learning methods of prediction and compare these approaches to existing methods, (2) to examine model-based approaches to combining multiple biomarkers for long-term or dynamic prediction, and (3) to develop and compare methods for examining changepoints in biomarkers. These aims complement the analytic approaches described in the Biostatistical Core for applying established methods to examine the data to be collected in the next funding cycle.