Mixed

Increasing the training sample size of DNA methylation-based epigenetic clocks improves prediction accuracy of chronological age, which attenuates the association between age acceleration residuals (AAR) and mortality, suggesting previous associations were confounded by cellular composition.

Current 'epigenetic clocks' (like Horvath or Hannum) that claim to measure 'biological age' and predict mortality may be flawed because they are confounded by blood cell counts. Using larger, more precise datasets to build these clocks removes the mortality prediction, suggesting these clocks measure chronological time accurately but do not necessarily predict healthspan or lifespan. Do not rely on standard AAR metrics for mortality risk assessment without adjusting for cellular composition.