Speech clocks reflect dementia phenotypes and biological aging markers

Researchers evaluated 2,928 individuals across five Latin American countries, including healthy controls and patients with mild cognitive impairment, Alzheimer's disease, and frontotemporal dementia. Using multimodal acoustic and linguistic features, the team trained supervised models to estimate chronological age and derived speech age gaps. These speech age gaps differentiated healthy individuals from patient groups, showing progressively higher values from Alzheimer's disease to non-language-dominant and language-dominant frontotemporal dementia. Speech gaps associated with clinical and cognitive domains, as well as the social exposome. In patients with Alzheimer's disease, speech gaps correlated with phosphorylated tau. They also associated with structural and functional brain clocks across dementia subtypes, and correlated with Hannum, Retroclock, and OMICmAge epigenetic clocks.