KDM: estimating an age-like latent state
The Klemera–Doubal Method was developed to combine multiple biomarkers that change with chronological age while accounting for each marker’s relationship with age and residual variability. The result is an estimate expressed in age units.
KDM is attractive for biological-age work because it provides an explicit statistical framework for combining biomarkers rather than simply averaging age predictions.
Levine PhenoAge: translating mortality risk into age units
Levine PhenoAge (clinical) takes a different route. It combines chronological age with nine routine clinical chemistry and hematology biomarkers to estimate mortality risk, then expresses that risk on an age-like scale.
A higher or lower Levine PhenoAge therefore has a direct relationship to the mortality-risk model from which it was derived. This differs from treating every age-like score as a generic measurement of the same latent construct.
Why the models can disagree
- They use different biomarker panels.
- They optimize different mathematical objectives.
- They were derived and validated in different ways.
- Their sensitivity to missing or changing biomarkers is different.
BioMIR keeps the clocks separate
BioMIR presents KDM and Levine PhenoAge (clinical) as a Clinical Long View rather than folding them into the daily Adaptive BioAge composite. That preserves their published interpretation and avoids implying that a daily wearable-derived change is mathematically equivalent to a clinical laboratory clock.
A useful validation question is whether longitudinal changes in a higher-frequency model show directional convergence with episodic clinical clocks when both are observed over sufficient time. That is a research target, not an established BioMIR claim.