COIOS
Tomorrow's world Item new preprint

Uncertainty in a biological age estimate predicts mortality independently of the age gap

A probabilistic framework applied to more than 450,000 UK Biobank participants derived both a biological age gap and a biological age uncertainty from the width of predicted-age distributions, with uncertainty only weakly correlated with the gap and independently associated with mortality.

Researchers modelled predicted-age distributions rather than point estimates in over 450,000 UK Biobank participants, separating the mean deviation (biological age gap) from the width of the distribution (biological age uncertainty). Uncertainty was weakly correlated with the gap and rose with disagreement between ageing signals across molecules and organs. Per standard deviation it carried mortality hazard ratios of 1.24 for NMR-derived and 1.40 for proteomic-derived measures, independent of the gap, with replication in three external cohorts using routine blood tests.

Why it is interesting: Treats the width of a predicted-age distribution as a predictor of mortality and healthspan loss separate from its mean.

Source
medRxiv, 28 September 2026
DOI
10.64898/2026.09.26.26364091
Type
Preprint
Design
Probabilistic biological-age modelling in 450,000+ UK Biobank participants with NMR, proteomic and routine-blood panels, prospective mortality and healthspan follow-up, GWAS, and replication in three external cohorts
Verdict
New finding
Driver
Slowing ageing itself
Driver
What linked data reveals