Prediction of brain age using structural magnetic resonance imaging: A comparison of clinical validity of publicly available software packages

Predicting brain age from structural magnetic resonance images is commonly used as a biomarker of biological aging and brain health. Ideally, as a clinically valid biomarker, brain age should indicate the current state of health and also be predictive of future disease onset and detrimental changes in brain biology. In this preregistered study, we evaluated and compared the clinical validity, i.e., diagnostic and prognostic abilities, of six publicly available brain age prediction packages using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Baseline brain age significantly differed between groups consisting of individuals with normal cognitive function, mild cognitive impairment, and Alzheimer’s Disease for all packages, but with comparable performance to gray matter volume. Further, estimates were not centered around zero for cognitively normal subjects and showed considerable variation between packages. Finally, brain age did not show strong correlation with disease onset, memory decline, nor gray matter atrophy within four years from baseline in individuals without neurodegenerative disease. The substantial bias and variability in cross-sectional predictions, combined with the weak associations between brain age and longitudinal changes in memory performance or grey matter volume, suggested limited validity of current brain age estimates as a biomarker for biological aging.