Abstract: Alzheimer’s disease (AD) is the major subtype of dementia, which accounts for 60-80% of dementia cases in the United States (US). With the increased life expectancy in developed countries, AD is becoming a major issue at the late stage of life. Heart disease and stroke events are another two growing issues among the elderly population. There is an expanding body of literature implicating heart disease and stroke as risk factors for AD, highlighting significant pathways at the cellular level. However, further research is needed on the association between heart, stroke, and development of AD to improve our understanding of epidemiological features of this link, which could lead us to identify high-risk subgroups who potentially disadvantaged from such dependencies. Hawaii's population is one of the most diverse ethnic populations in the US and there exist significant racial/ethnic health disparities in this multiethnic population. Especially, Native Hawaiians and Pacific Islanders (NHPI) are a well-known high-risk group for many disease conditions, including cardiovascular conditions and stroke. Links between heart dysfunction and stroke on developing AD makes NHPI potentially a high-risk group for AD. We propose a study to explore the AD risk for patients age 65 and above those who had major heart events and stroke, based on the Hawaii Medicare database. The proposed work will be conducted under a time-to-event multistate model framework while accounting for competing risks for acquiring unbiassed estimations and inferencing. Nine years of Hawaii Medicare data from 2009 to 2017 will be utilized to gather important time-to-event details for the proposed work, which allows us to conduct comprehensive time-to-event analyses on several aspects: progression from heart disease or stroke to AD; AD mortality; heart disease and stroke risks for AD. A key feature of the proposed study is the investigation of racial/ethnic dependency on AD, heart disease, and stroke links. We also propose to develop machine learning-driven predictive models for predicting transitions and mortality based on subject profiles.