ABSTRACT Alzheimer's disease and related dementias (AD/ADRD) are a group of progressive, terminal illnesses that will affect an estimated 14-million people in the United States by the year 2050. Caregivers experience chronic stress and they report feeling burdened and unprepared for making difficult end-of-life (EOL) care decisions, which may lead to unnecessary hospitalizations or avoidable transitions in care that affect end-of-life quality-of-life (EOL-QOL). Advance care planning (ACP), which improves EOL-QOL, is the gold-standard approach for improving concordance between preferences and actual care received at EOL. However, despite decades of research aimed at raising their rates, only 50% of those with AD/ADRD have a written ACP. Furthermore, there is a lack of current research evidence investigating the factors associated with transitions in care and EOL-QOL for persons with AD/ADRD, which could help guide EOL decision-making. To date, the state of the science is primarily cross-sectional in nature, and does not account for the influence of trajectories of decline, the effect these changes have on caregivers, nor how longitudinal changes in caregiving ultimately affect EOL care outcomes. Therefore, there is a critical need to discover new approaches for preparing persons with AD/ADRD and their caregivers in making informed, in-the-moment decisions, to ensure high EOL-QOL care and to support appropriate transitions in care as circumstances change over time. Using the National Health and Aging Trends Study (NHATS) and National Study on Caregiving (NSOC), we plan to use a machine learning based framework to identify the key determinants for predicting the risk for EOL care transitions and the traits of EOL-QOL among older adults residing in the community. This study has two specific aims: 1) Develop predictive model of factors related to end-of-life care transitions (e.g. inpatient death versus hospice) in persons with AD/ADRD longitudinally; and 2) Develop a predictive model of factors related to end-of-life quality-of-life (EOL-QOL) in persons with AD/ADRD. Discovering knowledge in a large population- level dataset is foundational for the future development of a generalizable/scalable model for guiding persons with AD/ADRD and their caregivers as they navigate a fragmented healthcare system while making difficult decisions for their loved ones. Our approach fills a critical gap between the current approaches for improving EOL-QOL and EOL transitions in care that focus on ACP as a singular outcome, by addressing the comprehensive needs of individuals with AD/ADRD and their caregivers that change over time. Our study will provide a predictive model for EOL-QOL and EOL care transitions. This is a critical first step for the future development of an approach for personalizing care to guide persons with AD/ADRD and their caregivers in making EOL care decisions. These results will have an important positive impact on EOL care, which aligns with ...