Speech age gaps track dementia phenotypes and biological aging

In a cross-sectional study of 2,928 human participants across five Latin American countries, researchers developed a speech-based biological aging clock. The cohort included healthy controls alongside individuals with mild cognitive impairment, Alzheimer's disease, and non-language or language-dominant frontotemporal dementia. Supervised models trained on multimodal acoustic and linguistic features estimated chronological age, yielding speech age gaps where positive values reflect older-appearing speech. Speech age gaps differentiated diagnostic groups, with healthy controls scoring lower than patient groups, and Alzheimer's disease scoring lower than frontotemporal dementia variants. These gaps associated with clinical and cognitive domains, structural and functional brain clocks, and social exposome. In Alzheimer's disease, speech age gaps correlated with phosphorylated tau (p-Tau217). The metric also correlated with epigenetic age clocks, including Retroclock and OMICmAge.