Development and validation of efficient cognitive composite scores of digital tools for the detection of early pathophysiological changes in Alzheimers disease

PROJECT SUMMARY Cognitive decline in preclinical Alzheimer’s disease (AD) has been closely correlated with tauopathy as measured by Positron Emission Tomography (PET). Other pathophysiological factors (e.g., amyloid-beta and neurovascular dysfunction) associated with cognitive decline contribute to the biological and clinical heterogeneity in preclinical AD. Identifying this heterogeneity is crucial for improving the detection of preclinical AD in cognitively normal (CN) individuals. Capturing pathophysiological events in preclinical AD has been challenging with traditional pencil-and-paper cognitive assessments. Features of a cognitive digital clock drawing test (dCDT) that capture time-stamped coordinates of drawing outcomes and process have demonstrated superior correlations with tau PET in elderly CN subjects. However, the dCDT features used in current studies were selected based on the clinical diagnosis between AD and CN subjects. The overall goal of this project is to 1) identify novel dCDT features most correlated with underlying AD- related pathophysiology measured using state-of-the-art multimodal neuroimaging techniques, accounting for biological and clinical heterogeneity in preclinical AD, and 2) establish the longitudinal performance of dCDT in capturing cognitive decline in preclinical AD. The identification of the novel dCDT features will provide a non- invasive assessment of cognitive status in the context of progressive pathophysiological events in preclinical AD. This will enable efficient and cost-effective participant screening for AD therapeutic trials, facilitate the translation of digital testing into AD primary care settings, and in doing so, improve its access for underserved populations. During the K99 phase, we will identify novel dCDT features most correlated with tau PET measures using machine learning and deep learning. To accomplish these goals, the applicant will leverage existing skills in PET imaging and machine learning and gain additional training in 1) multimodal neuroimaging, 2) deep learning, 3) pathophysiological, and 4) neuropsychological changes related to normal aging and AD. With these training activities, the applicant will be well poised to conduct the R00 phase to evaluate the dCDT performance in the context of clinical and biological heterogeneity in preclinical AD using multimodal neuroimaging techniques. This award will facilitate the applicant’s transition to an independent researcher who will utilize multimodal neuroimaging to better understand the biological mechanisms underlying brain disorders.