Dementia ascertainment is expensive and time consuming, as it typically involves an in-person study visit with cognitive assessment, neurologic exam, and functional assessment as well as an informant interview. Algorithmic dementia ascertainment — using cohort data or administrative data — is an attractive alternative, as it is easier, cheaper, and can be used to identify persons with dementia retrospectively. However, algorithmic ascertainment also has limitations, including real concerns about bias induced by use of algorithmic assessment. In this talk, Dr. Melinda Power will discuss her work to evaluate existing algorithms for dementia ascertainment, develop new algorithms for dementia ascertainment, and explore the potential for bias due to use of algorithmic ascertainment in the context of the U.S. nationally-representative Health and Retirement Study (HRS). She will close with recommendations on how best to use algorithmic ascertainment of dementia to answer important questions in dementia research.