Why I don't track my biological age

The question I'm asked most often this year is whether someone should buy a biological age test. The category has gone mainstream in 2026 — direct-to-consumer DNA methylation panels, telomere length tests, glycomics-based ageing scores, multi-omic clocks, wearable-derived "biological age" estimates. Most reasonable adults reading longevity content encounter the offering and feel they should probably try it.

I don't recommend it. Here's why, briefly, and what I think is worth doing instead.

What biological age testing actually is

Strip the marketing, and biological age tests are statistical models that take some biological measurement and predict chronological age in a reference population, then compare your reading to where you'd "land" on that prediction. The deviation between your predicted age and your actual age is your "biological age."

The original methodology — Steve Horvath's epigenetic clocks, published 2013 — was a methodological breakthrough. Methylation patterns at specific CpG sites correlate with chronological age across cohorts, and the prediction error of the clock has reproducible associations with mortality and disease risk in research settings. The science underlying the original Horvath clock is real.

The leap from "this is a meaningful research tool" to "this is a meaningful individual decision tool" is large, and the consumer market has not fully made it.

What the evidence does and doesn't support

What's reasonably supported in research settings:

• Population-level association between epigenetic age acceleration and mortality risk.
• Group-level changes in biological age estimates after long-term lifestyle interventions.
• Use of these metrics in research as outcome variables for studies of ageing.

What's much less supported:

• The reliability of any single individual measurement at a single time point.
• The actionability of small changes in the score from intervention to retest.
• The discriminative validity of the various consumer clocks (different products often disagree by years on the same person at the same time).
• The clinical decision value at the individual level outside research.

The honest summary: biological age scores capture something real at the population level, in research settings, with appropriate caveats. They are noisy at the individual level, expensive, and produce results that are difficult to act on in any way that the basics (sleep, movement, nutrition, stress practice, social connection) wouldn't already capture.

The specific concerns

Three concrete reasons I don't recommend the consumer category:

Test-retest reliability is worse than the marketing suggests. Repeat testing of the same person on the same day, with consumer products, often produces differences of multiple years. This is because the underlying methylation noise, sample handling, and statistical model variance all stack up. The number that comes back may be more about laboratory variance than your biology.

The actionability is unclear. Imagine your biological age comes back two years older than your chronological age. What do you do? More sleep, more exercise, less stress, better diet — but you'd be doing those things regardless. The score doesn't change the recommendation.

It anchors anxiety. I see patients become preoccupied with the score in ways that are not, on net, good for their healthspan. The wellness use case for biological age testing often defeats its own intent — adding chronic low-grade stress about a noisy number is unlikely to help anyone live longer.

What I'd track instead

If you want to track healthspan-relevant variables in a way that has more signal-to-noise than biological age testing, the unromantic answers:

• VO₂max, measured periodically (every 1–2 years). Probably the single most-replicated mortality predictor. Either a proper cardiopulmonary exercise test or a well-calibrated submaximal estimate.
• Resting heart rate trend over months. Cheap, available from any half-decent wearable, with a real signal for cardiovascular adaptation.
• Blood pressure, regularly. Boring, well-validated, modifiable.
• Standard lipid panel and HbA1c, periodically. The basics that drive most cardiovascular and metabolic risk. ApoB is the single most informative lipid measurement.
• Grip strength trends with age. Predictor of frailty and mortality, easy to measure.
• Self-reported sleep regularity, energy, mood across weeks. Subjective but actionable.

Five of these can be done at any GP appointment with no special technology. None of them require a $400 test that produces a number you don't quite know what to do with.

On the broader concept

The broader claim — that we should be measuring our biological ageing and intervening to slow it — is appealing in principle. The aspiration to extend healthspan is worth taking seriously. The current state of biological age testing isn't, in my view, the load-bearing way to engage with that aspiration.

The aspirational version of biological age testing — comprehensive, well-validated, clinically actionable, integrated with care — is a research project that may or may not become a consumer reality in the coming decade. The current consumer version is mostly a way of paying for anxiety.

I'll revisit this in five years and see whether the picture has changed. For now, I keep tracking sleep, energy, training metrics, and the unromantic clinical numbers; I don't track biological age; I don't think most people should.

This is also why we've committed at Healicus not to offer biological age scoring even when the underlying technology improves. The infrastructure to do it well, the data validation to make it trustworthy, the ongoing clinical context to make it actionable — these aren't the right tools for an educational platform. The honest version of healthspan support is the boring foundational work, sustained for years. That's what I'd point you at, even if the marketing of the rest is more exciting.

— Carmen