Lab Affiliation: Craig Stark Lab
Title: Screening for cognitive impairment due to Alzheimer’s using cost-effective data
Abstract: Access to geriatric care is important for managing Alzheimer’s and related dementias, but expert time and clinical resources are limited. To improve clinical efficiency and reduce costs, we aim to screen for cognitive impairment due to Alzheimer’s using lower-cost data modalities. We tackle this challenge from two perspectives: prediction and clustering. We train a random forest classifier to predict clinical diagnoses and, in parallel, cluster subjects into clinical subtypes. Using information obtainable from primary care visits, we can accurately predict diagnoses of mild cognitive impairment and dementia due to Alzheimer’s. Clinical subtype information further allows us to infer the clinical trajectory of subjects over time when we tap into longitudinal visit outcomes. This combined approach builds a holistic, cost-effective way to screen for Alzheimer’s disease.