InferAging: A non-invasive aging clock for quantifying individual differences in aging

Abstract

Quantifying differences in aging among individuals of the same chronological age could provide a direct measure of biological aging. However, existing methods often estimate biological age by predicting chronological age and treat such individual differences as prediction errors. We developed InferAging, a framework that estimates biological age by explicitly modeling individual deviations from chronological age. In progeroid zebrafish (klotho mutant; kl-/-), InferAging identified accelerated and delayed agers within the same chronological age group. A non-invasive variant using behavioral and morphological snapshots reproduced the transcriptome-integrated estimates without molecular input or lifelong tracking. The inferred aging state was associated with metabolic decline, intestinal barrier dysfunction, inflammation, and mucosal immune abnormalities beyond chronological age. These results demonstrate that non-invasive phenotypes can reveal molecularly supported individual aging states.