Predicting Cognitive Decline in Healthy Adults using Structural MRI

Abstract

Alzheimer’s disease (AD) is a progressive neurological disorder and the leading cause of dementia, affecting millions globally. As populations continue to age, AD prevalence is expected to rise significantly, placing substantial burdens on healthcare systems. Early detection of cognitive decline is essential, as interventions during preclinical stages may slow disease progression and improve quality of life. In this study, we analyzed structural MRI (sMRI) data from 194 participants aged 70 and above from UK Biobank, focusing on specific brain features to identify predictors of cognitive decline. Using machine learning techniques, we assessed a diverse set of sMRI features, including regional grey matter and subcortical volumes, gray-white matter intensity ratios, multiple parcellations of the white surface, and a parcellation of the pial surface, capturing distinct aspects of brain structure. Each feature group included multiple measurements representing different dimensions of brain morphology. We evaluated the predictive performance of each feature group to determine which provided the most accurate prediction of cognitive decline. Cognitive performance was assessed through Fluid Intelligence (FI) and Reaction Time (RT) tests, allowing classification of participants as “Cognitive Decliners” vs. “Non-Decliners”. Our findings underscore the predictive value of these sMRI features in identifying early indicators of cognitive decline, advancing the development of MRI-based biomarkers to support timely interventions for Alzheimer’s and related dementias.