Assessing the Utility of Predicted Brain Age for Explaining Variability in Language Abilities in Healthy Older Adults.

We investigated whether the difference between chronological and modeled brain age explains individual differences in language performance among healthy older adults. Age-related decline in language abilities is widely documented, with considerable variability among healthy older individuals in both language performance and underlying neural substrate. We derived predicted brain age from grey and white matter using machine learning and used this measure to estimate neurological deviations from chronological age. Using Bayesian mixed-effects modeling, we tested whether brain-age deviations predict language performance in a sample of 86 adults aged 60 years and above. We assessed the effect of brain-age deviations on performance across four well-established language processing tasks, each tapping into linguistic domains known to be vulnerable to ageing and show individual variability in skill levels, in both comprehension and production. Our findings suggest that, in healthy older individuals, predicted deviations of brain age from chronological age do not predict language abilities. This challenges the idea that brain age is a reliable determinant of language processing variability, at least in healthy (as opposed to pathological) ageing and highlights the need to consider other neural and cognitive factors when studying language decline.