Psychiatry & Human Behavior logo Click here for the Department of Psychiatry & Human Behavior Home Page Click here for the UC Irvine Home Page Click here for the School of Medicine Home Page

Predicting Dementia in Adults with Down Syndrome

David B. Keator M.S.  PhD
Francisco Chanes
David B. Keator, M.S., Ph.D. Operations Director, Neuroscience Imaging Center

UCI Researcher uses PET to predict dementia in adults with Down Syndrome

Dr. David Keator's education consists of a B.S. inBiological Sciences from UCI, an M.S. in Computer Science from California State University, Long Beach specializing in software engineering, and a Ph.D. in Computer Science from UCI specializing in machine learning. Keator is currently the Operations Director of the Neuroscience Imaging Center (NIC) and an Associate Professional Researcher.

Dr. Keator is engaged in a multi- faceted research program focused in three principle domains: (1) Identification of biomarkers and neuro-phenotypes for psychiatric disorders; (2) Development of machine learning models for problems in neuroimaging; (3) Developing models and techniques in knowledge representation, management, and retrieval for problems in medicine. In each of these domains Dr. Keator has made significant research contributions, many of which have gained international exposure.

Examples include contributions to improving data acquisition quality using a probabilistic graphical modeling approach to improve system tuning in PET, building algorithms to detect functional abnormalities in brain imaging data based on hierarchical models of the ventral visual stream, and the development of the Neuroimaging Data Model (NIDM), a next-generation, semantically-annotated, graph-based metadata format for neuroimaging.

Dr. Keator has been working on brain- based biomarkers for early dementia prediction in Down Syndrome with Dr. Ira Lott. Down syndrome (DS) is associated with elevated risk for Alzheimer’s disease (AD) and lifelong accumulation of beta amyloid (Aβ). He hypothesized that the spatial distribution of Aβ plaque burden predicts transition to dementia in individuals with DS. We acquired 18F- Florbetapir PET scans from 19 non- demented adult individuals with DS at baseline and monitored them over a four-year period, identifying 5 individuals who transitioned to dementia. He used machine learning classification to determine features on 18F-Florbetapir maps that were most predictive of transition. This work resulted in showing that regional spatial patterns of amyloid accumulation could be used to accurately predict which participants would develop clinical dementia in the future. In addition to “AD signature” regions including the inferior parietal cortex, temporal lobes, and the cingulum, we found that Aβ cortical binding in the orbitofrontal and inferior parietal cortices distinguished subjects who transitioned to dementia from those who did not.  This work was recognized by the international community and invited for an oral presentation at the 2017 Alzheimer’s and Parkinson’s Disease conference. Dr. Keator continues this work with Dr. Lott and others as part of a large U01 grant to develop broad-based, multi- modal risk scores for early dementia prediction.

Dr. Keator has many hobbies he shares with his wife and son. He is an Assistant Scoutmaster for his son’s Boy Scout troop, he is an avid surfer and skier, he is a guitar player who became a luthier and now hand makes acoustic guitars, and an amateur car racer who placed third in his class at a recent event.