Biological age: how epigenetic clocks are calculated

A "biological age" test returns a number that can be compared to your chronological age. The difference between them is presented as the result of your lifestyle. In reality, the number is the output of a statistical model trained to predict something specific, and what that difference means depends entirely on what the model was trained to predict.

How it works

DNA methylation is the addition of methyl groups to specific regions of the genome. The pattern of methylation changes predictably with age, and the "clocks" are built on this: they take several hundred sites, train a regression to predict age, and produce a prediction in years.

The key point: the clocks do not measure age. They predict the value they were trained on. The first versions were trained on chronological age, and they predicted it well—meaning, by definition, they could not tell you anything beyond what you already know.

What subsequent versions changed

Later, models began to be trained not on age, but on outcomes. GrimAge was trained on mortality and biomarkers associated with mortality, including a surrogate for smoking history. The developers describe it as a biomarker that predicts life expectancy and time to disease better than previous versions. Publication in Aging.

This changes the meaning significantly: a value that correlates with mortality in a population is not the same as an indicator reflecting the state of a specific individual. A review of biological age predictors analyzes different approaches and their limitations — EBioMedicine.

Why a repeat test gives a different number

A separate and little-known problem is technical reliability. A study in Nature Aging, dedicated specifically to this, describes the low reliability of repeated measurements in a number of clocks and proposes a computational solution, while specifically noting the implications for clinical trials and longitudinal tracking. Publication in Nature Aging.

A difference of two or three years between two tests can fall entirely within the margin of error of the method. Yet, this is exactly the difference sold as a protocol result.

Added to this is a biological source of noise. Methylation is measured in blood, and the composition of blood by cell type changes due to infection, inflammation, or time of day. Some models adjust for cell composition, others do not.

What "three years younger" means

This phrase usually means: the model output was three years lower the second time than the first. It does not follow that the state of your tissues has changed, nor that your risk of disease has changed.

For such a conclusion to be valid, a trial is needed where an intervention changes both the clock indicator and clinical outcomes, and in the same direction. This is exactly the gap between a surrogate endpoint and an outcome that is discussed in the article on anti-aging supplements.

What laboratories do with this

A commercial test usually does not disclose which specific clock is used, what the model was trained on, or what the stated reproducibility is. Without these three points, the result cannot be verified or compared with the result from another laboratory.

The situation is similar with home tests in general: the answer looks precise because it is expressed as a number. More on this in other categories — in the articles on DNA tests and nutrition and on microbiome tests.

Where it is actually used

Epigenetic clocks are a working tool for population studies. There, they are applied to groups where random noise is averaged out, rather than to a single person, where it determines the result. The distinction between "the indicator works on a sample of a thousand people" and "the indicator tells me something about myself" is central to the entire topic of self-measurement; see the article on self-experimentation.

Count it from a photo

Frequently asked questions

What do epigenetic clocks measure?

The pattern of DNA methylation in several hundred regions of the genome. A statistical model then converts this into a number of years. The clocks do not measure age; they predict the value they were trained on.

How does GrimAge differ from early clocks?

Early versions were trained to predict chronological age. GrimAge was trained on mortality and mortality-related biomarkers, which is why it predicts life expectancy and time to disease onset better than previous versions.

Why do two consecutive tests give different numbers?

A number of clocks have low test-retest reliability—a topic covered in a separate study in Nature Aging. Biological noise is also added: methylation is measured in blood, and its cellular composition changes during infections, inflammation, and throughout the day.

What does it mean that my biological age has decreased by three years?

It means the model output was three years lower the second time. This does not imply a change in tissue condition or disease risk, and the difference itself may fall within the method's margin of error.

Is it possible to compare results from two different laboratories?

Usually, no. Commercial tests do not disclose which clocks are used, what the model was trained on, or what the stated reproducibility is.

Read next

• Longevity supplements: NMN, resveratrol and what human trials measured — This shelf has the best story of any: a molecule, ageing cells, mice that lived longer.
• DNA test "tailored to your genome": what is known about nutrition based on genes — The promise sounds convincing: a saliva sample, two weeks of waiting, and a diet tailored to your heredity.
• Microbiome tests: why two samples give different answers — An at-home microbiome test looks like a standard lab analysis: a test tube, a laboratory, and a report with percentages and a list of foods.

This article is for general information. It is not medical advice, a diagnosis, or a prescription for treatment or a diet, and it does not replace a consultation with your doctor. If you have a health condition, are pregnant, take medication, or follow a diet prescribed to you, decisions about food belong with your doctor.

Figures from regulations, guidelines and studies are given as they stood when this article was prepared and may since have changed; check them against the primary sources. This article is not advertising, an offer, or individual advice, and neither the author nor the site owner is responsible for decisions taken on the basis of it.