Dementia-Detecting AI Tool Wins National Institute on Aging Prizes

Assistant Professor Maryam Zolnoori, PhD, and her team have received two national honors from the National Institute on Aging’s PREPARE Challenge for their groundbreaking work using speech to detect early cognitive impairment. 

They received a Special Recognition Prize for Algorithm Development and the Explainability Bonus Prize for creating a model that not only predicts Alzheimer’s risk but also offers clinically interpretable explanations to support health care decision-making. 

Their award-winning work was part of Phase 2 of the PREPARE: Pioneering Research for Early Prediction of Alzheimer’s and Related Dementias EUREKA Challenge, a national competition focused on advancing “accurate, innovative, and representative” tools for early detection of Alzheimer’s disease and related dementias (ADRD). Competing teams worked in one of two tracks: the Acoustic Track, which uses speech and voice recordings, and the Social Determinants of Health Track, which analyzes longitudinal data on social and environmental factors. 

ADRD affects one in five adults over 60, yet more than half remain undiagnosed and untreated—leading to serious declines in quality of life. As new disease-modifying therapies emerge, early identification is more critical than ever. 

Zolnoori’s algorithm analyzes voice recordings captured during routine patient-clinician communication. “This technology is cost-effective, non-invasive, and remarkably sensitive,” she explains. “Because it works with everyday clinical interactions, it can be seamlessly integrated into health care workflows to detect cognitive decline early.” 

Building on this success, her team is now leading a 300-patient, NIA-funded clinical trial of a speech-based screening tool designed for home health care settings. More than 100 participants have already been enrolled.