Aging Clocks Need the Right Reference: What Sex-Specific Models Show

A report says your liver looks older than your birthday. Before asking how to make the number smaller, there is a more basic question: older compared with whom?

A new Nature Medicine paper addresses that reference problem. The investigators developed 38 sex-specific clocks across 15 organ systems using imaging, proteins and metabolites. They reported sex- and organ-dependent genetic, molecular and clinical associations. These are research models, not evidence that a supplement reverses organ aging. Nature Medicine, September 16, 2026.

What an organ age gap actually means

A biological age gap compares a model’s predicted age with chronological age. Its meaning depends on the data and reference group used to train the model. It should not be read as a direct measurement of how many years an organ has left.

Consider a deliberately simplified example, not a result from this paper: if a model predicts age 65 for a person aged 60, the arithmetic gap is five years. That does not mean five years of life have been lost. Changing the reference model could change the estimate without changing the person at all.

This is why prediction accuracy, calibration and clinical usefulness need separate evaluation. A model can predict a birthday reasonably well yet still fail to tell a clinician whether a particular intervention will help.

Different references do not create a faster-aging winner

The study supports examining sex-specific references while retaining pooled and sex-interaction approaches as complementary tools. It does not justify ranking all women against all men on one universal aging scale. Original paper, Discussion.

Two independently trained clocks can use different feature weights and reference distributions. A difference between their scores can therefore reflect both biology and the way the ruler was constructed. The useful question is whether each ruler predicts outcomes reliably for the people to whom it is applied.

Association is not a pathway switch

A protein associated with an age-gap score is not automatically a cause of aging. Nor does a positive association demonstrate that an entire pathway was experimentally switched on. Negative associations do not, by themselves, demonstrate beneficial downregulation.

For a mechanism illustration, the honest starting point is a measurement diagram: imaging or molecular measurements enter a model; the model yields an estimate; researchers examine relationships with other measurements and outcomes. Causal arrows require evidence beyond a correlation. Genetic-instrument analyses also depend on assumptions about the instruments and alternative routes to the outcome.

We therefore use conceptual organ and clock imagery, not invented molecular structures, treatment arrows or a fabricated before-and-after rejuvenation chart.

The limits matter for real people

The authors identify restricted repeat measurements, European-ancestry genetic analyses, binary genetically proxied sex, and largely sex-pooled disease-genetics endpoints as limitations. They also call for external validation at biobank scale. These details constrain generalization and claims about within-person aging rates. Original paper, Limitations.

For a consumer test, ask whether the exact assay and model were evaluated in people like you, how repeatable the result is, and whether acting on it improves outcomes. Validation of one research clock does not automatically validate a commercial test with a similar name. This paper should not be used to choose hormones, supplements or medications from a clock score.

Changing a clock remains a different question

Separate contextual evidence comes from the CALERIE randomized trial analysis in Nature Aging (2023): a calorie-restriction intervention affected DunedinPACE but did not significantly change PhenoAge or GrimAge. This is not a validation of the new sex-specific organ models; it illustrates that different clocks need not respond together. PMID 37118425.

For Magellan’s product scanner, this is research on biomarker modeling, not a product-efficacy claim. A smaller score after taking a product would still require careful interpretation rather than an automatic anti-aging endorsement.

The promising idea is a better comparison. The unanswered question is whether using that comparison changes care and improves health.

Source note: The main paper is linked by its verified publisher DOI. A matching PubMed record was not found during this review; no PMID has been assigned to it here. The PubMed-linked CALERIE paper is separate contextual evidence. This analysis draws on the main paper’s published results, methods and limitations; it does not reproduce the investigators’ calculations.

Educational, not medical advice.