Women’s and men’s organs just got separate aging clocks

On 16 September 2026, Nature Medicine published 38 sex-specific biological aging clocks spanning 15 organ systems, built by Zhiyuan Song, Junhao Wen, and the MULTI Consortium at Columbia University and partners — and the clocks do not treat female and male aging as the same curve.

The hedge belongs up front: these are AI age-prediction gaps (predicted age minus calendar age), not a pill that adds years. Performance sits near a ~5-year mean absolute error on hold-out tests — useful for research and risk stratification, not a bathroom-mirror oracle. Genetic analyses lean on European-ancestry UK Biobank data, and many disease GWASs used in Mendelian randomization remain sex-pooled. Still, when the same organ is scored against a female-only versus male-only healthy reference, disease and mortality links rearrange in ways pooled clocks can blur.

Why it matters

Women live longer on average yet often carry more late-life morbidity; men tend to hit cardiometabolic disease earlier. Alzheimer’s disease shows sex-skewed burden and treatment response. Most “biological age” products and papers still train on mixed-sex samples and quietly assume the math is sex-invariant.

If your brain, heart, or liver is judged against the wrong sex’s normal curve, a “young” or “old” label can be noise. For trialists using organ clocks as enrichment biomarkers, that noise becomes who gets enrolled. The stake for a stranger is concrete: better risk tools for hypertension, diabetes, and the slide from mild cognitive impairment (MCI) into Alzheimer’s — especially if those risks are not the same size in women and men.

Key numbers

What happened

Wen’s group had already shipped multi-organ clocks from MRI, proteomics, and metabolomics — but those earlier clocks were sex-pooled. Here they retrained separately in healthy women and healthy men, using nested cross-validation and matched sample-size checks so the comparison is fair.

On a within-distribution hold-out set, sex-stratified models performed in the same ballpark as prior pooled clocks (MAE ~5 years). Some organs still overfit; the team kept 18 male and 20 female clocks that cleared their overfitting screen for downstream work. Comparing biological age gaps (BAGs) between sexes showed significant differences for most organ pairs they tested — not because one sex is universally “older,” but because the reference trajectory itself splits.

Genetics followed. Sex-stratified GWASs produced hundreds of locus–BAG links. SNP-based heritability was often similar in magnitude yet organ-specific in which sex ran higher: hepatic proteomic, liver MRI, and metabolic metabolomic clocks looked more heritable in females; immune metabolomic, skin proteomic, and spleen MRI clocks leaned male. Polygenicity flipped too — endocrine proteomic aging looked far more polygenic in females; immune proteomic aging more polygenic in males.

Disease, death, and Alzheimer’s

Two decades of UK Biobank inpatient follow-up let the team ask whether sex-specific BAGs forecast future disease. In women, a tighter set of clocks linked to hypertension, hypercholesterolemia, type 2 diabetes, and — strikingly — bipolar affective disorder via a metabolic clock. In men, the map was broader: metabolic and endocrine clocks coupled to hypertension, lipids, and diabetes phenotypes, with additional heart, adipose, kidney, and lung signals.

All-cause mortality showed 12 significant BAG–mortality pairs in females and 14 in males, mostly risk-increasing across multi-organ proteomic and metabolic axes (with a few protective-looking MRI associations in men that the authors flag for cautious reading).

The Alzheimer’s case study is the phone-readable stake. Using ADNI, sex-specific brain MRI clocks did not clearly predict cognitively normal → MCI. They did predict MCI → AD in both sexes, and the hazard was larger in women: roughly +74% risk per standard-deviation older brain in men versus more than double the hazard in women (HR 2.25). In the preclinical A4 solanezumab trial, among people whose brains looked younger than calendar age at baseline, women held higher Preclinical Alzheimer Cognitive Composite (PACC) scores than men at week 240; that sex gap did not show in the accelerated-brain-aging stratum.

Molecular texture (without overclaiming)

Protein-wide associations tied MRI organ clocks to thousands of plasma proteins with sex-resolved signatures — for example, a female-specific brain–SLITRK1 link versus a broader male brain-protein set including GDF15. Male spleen aging associated with dozens of inflammatory markers including TNF; female spleen links were narrower. Those maps are association science, not drug targets yet, but they show why a single “inflammation clock” for everyone can miss the plot.

What this is not

• Not a consumer longevity app. MAE ~5 years and research cohorts are not a personal diagnosis.
• Not proof that pooled clocks are useless. The authors explicitly keep sex-pooled and sex-interaction models as complementary tools when biology is shared or samples are small.
• Not a full accounting of gender, hormones, or pregnancy. Binary genetically proxied sex; menopause and hormone therapy are only partly explored in secondary analyses.
• Not global ancestry coverage. External multi-ancestry validation is still needed.

What to watch

1. Sex-stratified disease GWASs from FinnGen / PGC so Mendelian randomization is not half sex-blind.
2. Prospective trials that enroll or stratify on sex-specific brain or heart clocks.
3. Whether geroprotectors shift female and male organ clocks differently — the same drug, two reference curves.
4. Open MEDICINE portal reuse: can outside labs reproduce the MCI→AD hazard split?

Sources

1. Song, Z., Feng, D., Rahman, N.A. et al. / MULTI Consortium (2026). Sex-specific biological aging clocks across organs and omics. Nature Medicine. DOI: 10.1038/s41591-026-04662-6. Published 16 Sep 2026.
2. News-Medical (Pooja Toshniwal Paharia), 17 Sep 2026 — Sex-specific biological clocks reveal how aging differs across the body in women and men.
3. MEDICINE portal (GWAS / clocks): https://labs-laboratory.com/medicine/

Author: The Good Signal. Cover: stock MRI/brain research imagery for editorial use.