PROJECT SUMMARY/ABSTRACT Currently, more than 6 million Americans are suffering from clinical Alzheimer's disease and Alzheimer's related dementias (AD/ADRD). By 2050, this number is projected to rise to 13 million. Due to this expected increase, there is a great need for sensitive, easily implemented, objective, and scalable methods to identify cognitive change and differentiate individuals along the AD continuum. Neuropsychological assessments are useful for characterizing dementia symptoms, clinical diagnosis, and monitoring, but they involve burdensome procedures and lack ecological validity. Neuroimaging and blood markers of proteins associated with risk of clinical AD/ADRD are often inaccessible, expensive and/or invasive, and their link to patient behavior is inconsistent. To address these drawbacks, digital phenotyping is a novel approach that allows for passive and continuous monitoring of behaviors and physiological metrics in everyday life using technology such as a smartphone or smartwatch. In this study, the construct validity of smartwatch-derived metrics will be investigated against conventional neuropsychological measures and questionnaires across preclinical AD/ADRD stages. A diverse sample of 90 participants aged 55 and older with either healthy cognition or mild cognitive impairment will be recruited from the Philadelphia region. Participants will be given a Garmin Vivosmart 4 smartwatch and will wear it 23 hours/day to passively monitor sleep, heart rate variability, and physical activity data, which then will be deidentified and transferred to a secure server daily for 30 days. A daily survey will be completed through study software that is downloaded on participants' smartphones to contextualize passive data collection metrics. At the first study visit, participants will complete conventional neuropsychological tests and self-report measures of everyday function. The primary aim will examine the validity of a smartwatch-derived digital phenotyping protocol against conventional clinical measures. Exploratory aims will investigate 1) associations between smartwatch metrics and demographic and contextual factors and 2) meaningful groupings of participants with unique patterns of smartwatch metrics (i.e., phenotypes) and their relations to clinical status and conventional cognitive measures. The proposed training plan was developed with input from a team of interdisciplinary experts in everyday cognition in aging, digital ethics, longitudinal statistical methods, and engineers studying physiology with the goal of developing a high level of competence in statistical methods for wearable devices, understanding physiological measurement, knowledge on ethics and privacy in the wearable space, and the application of digital phenotyping in a diverse aging population.