PROJECT SUMMARY / ABSTRACT Depression has an enormous impact: it is the leading cause of disability worldwide and affects more than 264 million people of all ages. Moreover, major depression at any age doubles the risk of Alzheimer’s Disease and related dementias, which affect over 50 million people worldwide — a number projected to triple by 2050. Interventions for depression in older adults have limited efficacy to date, and directly treating cognitive impairment and/or Alzheimer’s Disease is not feasible or effective. Thus, identifying modifiable risk factors that favorably influence both depression and cognition before dementia onset is an urgent public health need. Sleep is one such risk factor. However, sleep is not a uni-dimensional construct represented by merely its duration or the presence/absence of a sleep disorder. Rather sleep is multidimensional: it is comprised of multiple domains (e.g., Regularity, Satisfaction, Sleepiness, Timing, Efficiency, Duration) and measured on multiple levels (e.g. self-report or behavioral [via actigraphy]). In our initial R01, we leveraged our biostatistical and sleep expertise to develop and hone methods for examining multidimensional sleep health as a predictor of mortality in a high- dimensional machine learning (ML) context that flexibly accounts for the complex interactions that exist among sleep and non-sleep risk factors. We now seek to build on the success of the initial funding period by using our novel methods to examine multidimensional sleep health as a predictor of changes in cognition and depressive symptoms. To enhance generalizability and power, we are developing a Pooled Sample of N~3,400 adults aged ≥65 without cognitive impairment from the Osteoporotic Fractures in Men Study, Study of Osteoporotic Fractures, Memory and Aging Project (MAP) and Minority Aging Research Study (MARS). With these methods and data, we will examine multidimensional sleep health for predicting changes in global cognition and incident dementia (Aim 1) and depressive symptoms (Aim 2) in a high-dimensional machine learning context. We will also examine depression as a pathway through which multidimensional sleep health predicts impaired cognition (Aim 3). Our Secondary Aims are to: (a) apply parallel methods in two additional cohorts (the Rotterdam Study and Multi- Ethnic Study of Atherosclerosis) to replicate and extend our findings to cohorts with different demographic profiles and clinical Alzheimer’s Disease and related dementias diagnoses; (b) examine effects by sex and race; and (c) identify the sleep health characteristics driving overall effects. Identifying multidimensional sleep health profiles that reliably predict changes in global cognition, incident dementia, and changes in depressive symptoms in a realistic, high dimensional context will directly inform the design of novel targeted interventions and prospective studies focused on preventing Alzheimer’s Disease and related dementias. Moreover, we w...