Blog
Biological age explained: what the number can—and cannot—tell you
Biological-age scores summarise patterns in health biomarkers. They are estimates produced by a particular algorithm, not a literal measurement of how old your body is, and two tests can legitimately disagree about the same person.
Chronological age is fixed by the calendar. Biological-age models combine measurable features of physiology or molecular biology to estimate where a person sits relative to expected patterns of ageing and health risk. The estimate is only as meaningful as the inputs behind it.
Biological age is a model, not a diagnosis
The phrase sounds as though one test can reveal the true age of the body. In practice there are several different biological-age models and they do not measure the same thing. Some use routine blood biomarkers, some analyse DNA methylation patterns, and others estimate the rate of change over time rather than a single point.
Each is an algorithm trained on a population dataset to predict a particular outcome — often mortality or the onset of age-related disease. The output is expressed in years because years are intuitive, not because the model measured years.
A score matters most when you understand which inputs produced it and whether any of those inputs are modifiable health risks. A number with no component breakdown is close to useless clinically.
This distinction matters commercially as well as scientifically. A result reported as a single headline figure invites the response “how do I lower it?” — which is the wrong question if nobody has established what is driving it.
Different clocks answer different questions
Phenotypic models
Tools such as PhenoAge combine chronological age with laboratory markers associated with mortality and health risk. Useful for risk stratification, and every input is something a clinician can already act on.
Epigenetic clocks
These analyse DNA methylation patterns at specific sites. Different clocks are trained on different outcomes, so two can produce different estimates from the same sample.
Pace-of-ageing measures
Rather than estimating a current age, these estimate how fast a person is ageing over time. They answer a different question and are not directly comparable to a point estimate.
A result should always identify the model used. Comparing scores from different algorithms as though they were interchangeable is like comparing two exam results without knowing the subjects. The number is in the same units; it is not measuring the same thing.
There is also a practical difference in what you can do with each. Phenotypic models are built from markers that already have treatment pathways — blood pressure, glucose, lipids, inflammation, kidney and liver function. Epigenetic clocks may be biologically closer to the ageing process itself, but a methylation pattern is not something you treat directly.
Why two tests can give you different answers
People are often unsettled to find one test reporting them three years younger than their age and another four years older. This is usually not an error by either laboratory. It reflects genuine differences in method, and knowing the reasons makes the results more useful rather than less.
| Source of difference | What is actually happening |
|---|---|
| Different training outcome | One clock was trained to predict mortality, another to predict disease onset or physical function. They are optimised for different targets. |
| Different inputs | A blood-biomarker model and a methylation model are reading entirely different biology. |
| Biological variation | Recent illness, dehydration, acute inflammation, intense exercise in the preceding days and even sleep loss can shift biomarkers. |
| Laboratory variation | Assay platform, reference ranges and sample handling all introduce measurement noise. |
| Version differences | Clocks are periodically retrained. The same brand of test may not use the same algorithm it used two years ago. |
The practical consequence is that the direction of travel within one consistent method is more informative than the absolute figure from any single test. A result that moved from +4 to +1 using the same assay and the same algorithm tells you more than two different tests disagreeing by five years.
Acute infection, recent surgery, a flare of an inflammatory condition or significant recent weight change can all distort biomarker-based scores. A result taken in that window may reflect a temporary state rather than any underlying trajectory, and repeating it when well is usually more informative than acting on it.
What the result is genuinely useful for
For most patients the practical value is not the headline number. It is the pattern underneath it: blood pressure, glucose regulation, lipids, kidney and liver function, inflammatory markers, body composition, sleep quality, smoking status and physical fitness.
If a biological-age score is higher than expected, the next question is which conventional health risks are contributing. Those are the things with established interventions and decades of outcome data behind them.
There is a second, less discussed benefit. A single unexpected number can motivate a change that years of general advice did not. That is a legitimate use, provided the change being motivated is an evidence-based one — treating blood pressure, addressing sleep apnoea, resuming exercise — rather than a supplement protocol chosen because it claims to lower the score.
Related Aquila service
Our biological age and longevity assessment combines broad biomarker review with physician interpretation, rather than treating the age score as a stand-alone answer. Where the underlying question is conventional risk, our health screening pages set out what standard screening covers.
What tends to move the markers
| Change | What it typically affects | Comment |
|---|---|---|
| Blood pressure control | Cardiovascular and renal markers | Among the best-evidenced interventions in medicine, independent of any age score. |
| Glucose regulation | Fasting glucose, HbA1c, inflammatory markers | Often the largest single contributor when a score is unexpectedly high. |
| Regular aerobic and resistance exercise | Fitness, body composition, glucose handling, lipids | Resistance work matters for muscle mass, which is frequently overlooked. |
| Stopping smoking | Multiple markers across models | Appears in most models because it is one of the strongest mortality predictors. |
| Sleep and sleep apnoea treatment | Blood pressure, glucose, inflammation | Untreated sleep apnoea can undermine otherwise good control. See sleep apnoea assessment. |
| Diet quality and weight | Lipids, glucose, liver markers, inflammation | Sustained change matters more than short-term shifts. See weight management. |
| Treating existing conditions | Whichever markers that condition affects | Treating the disease is more productive than treating the score. |
These are the same fundamentals that improve conventional health outcomes. That is not a coincidence — the models were built from those markers because those markers predict outcomes. A longevity programme that distracts from them is working against its own premise.
It is possible to optimise a number without improving health, particularly where a protocol targets a marker directly. The reverse — improving health without the score following — is far less concerning.
What biological-age testing does not replace
It does not detect cancer, image the coronary arteries, assess bone density or replace age-appropriate screening. A favourable score offers no reassurance about any condition it was not designed to detect, and treating it as a general all-clear is the most consequential mistake people make with these tests.
Recommended screening — cervical, breast, colorectal and others according to age, sex and risk factors — runs on its own schedule regardless of what an age estimate says. Where earlier detection is the actual question, that is a different category of test entirely; see multi-cancer early detection for what those tests do and do not establish.
Nor does a single number predict an individual lifespan. These are population-derived estimates. They describe how a group with similar markers has fared, which is genuinely useful information and still not a forecast for one person.
Before you buy a biological-age test
Direct-to-consumer testing varies enormously in quality and in what you receive. A few questions separate a useful purchase from an expensive number.
| Ask | Why it matters |
|---|---|
| Which algorithm or clock is used? | If the provider will not say, the result cannot be compared with anything, including your own future tests. |
| Will I receive the underlying biomarker values? | The components are the clinically actionable part. A score alone cannot be acted on. |
| Is a clinician interpreting it? | An abnormal input may need investigation in its own right, independent of the age estimate. |
| What is the reported variability? | A test with wide test-retest variation cannot detect the small changes people hope to track. |
| Is a treatment programme bundled with it? | Where the same provider sells both the diagnosis and the remedy, the incentive is worth being aware of. |
The last point is not an accusation of bad faith, and it applies to clinics as much as to online sellers. It is simply a structural feature of the market worth recognising when a result is followed immediately by a recommendation to purchase something.
Repeat testing is most useful when something has changed
Repeating the same validated method after a meaningful interval can show whether the underlying biomarkers are moving favourably. Frequent retesting over short intervals is more likely to capture noise than meaningful ageing biology.
A reasonable trigger for repeating is a substantive change: treatment started for blood pressure or glucose, significant weight change, smoking cessation, treatment of sleep apnoea, or a sustained exercise programme. Retesting at three-month intervals with nothing changed in between mostly measures assay variation.
Use the same laboratory conditions and the same algorithm where possible, and review the individual components rather than only the final estimate. If the score improved but an individual marker worsened, that marker is the more important finding.
Questions worth asking about your own result
- Which inputs pushed my score in each direction? This is the actionable content.
- Is any individual result abnormal enough to need investigation on its own?
- Does anything here change my conventional risk assessment for cardiovascular disease or diabetes?
- What would we expect to change, and over what timeframe, if I address the main contributor?
- Am I up to date on the screening appropriate for my age and risk factors?
- Was I well when the sample was taken? If not, the result may need repeating.
If the answer to the first question is not available, the test has given you a number rather than information.
Questions patients ask
What is biological age?
An estimate produced by an algorithm that combines health biomarkers or molecular measurements to describe where someone sits relative to population patterns of ageing and risk. It is a model output, not a direct measurement.
Is biological age scientifically valid?
The underlying models are real research tools with published validation against outcomes such as mortality. That is different from any one commercial test being validated, which depends on the specific method used.
Why did two tests give me different ages?
Because they use different inputs, are trained on different outcomes, and carry biological and laboratory variation. Different answers from different methods are expected rather than an error.
Can biological age go down?
The estimate can fall if the markers behind it improve, typically through blood pressure, glucose, fitness, weight, smoking and sleep. Whether that reflects reversed ageing or improved risk profile is a separate question.
How often should I retest?
When something meaningful has changed — a treatment started, sustained weight or fitness change, sleep apnoea treated. Frequent short-interval retesting mostly measures noise.
Does a good score mean I do not need health screening?
No. It does not detect cancer, image the heart or assess bone density. Age-appropriate screening continues on its own schedule regardless of the score.
Does biological age predict how long I will live?
No. These are population-derived estimates describing how similar groups have fared. They are not an individual forecast, in either direction.
Which test is best, blood-based or epigenetic?
They answer different questions. Blood-biomarker models produce inputs a clinician can act on directly; epigenetic clocks may sit closer to ageing biology but methylation is not something treated directly.
Should I test while I am unwell?
Preferably not. Acute infection, recent surgery, inflammation and significant recent weight change can distort results, so a test taken then may reflect a temporary state.
What should I ask a provider before buying a test?
Which algorithm is used, whether you receive the underlying biomarker values, whether a clinician interprets it, the reported test-retest variability, and whether a treatment programme is bundled with the result.
Do supplements lower biological age?
Marketing claims are common and generally outrun the evidence. Improving the inputs the model reads — blood pressure, glucose, fitness, sleep, smoking — has the stronger case, and optimising a marker is not the same as improving health.
What is the single most useful part of the result?
The component breakdown. An abnormal individual marker may warrant investigation in its own right, whatever the composite score turned out to be.
Sources & further reading
This article is general educational information and does not replace individual medical assessment. Biological-age estimates vary by method, algorithm version and laboratory, and should be interpreted alongside established clinical risk factors and age-appropriate screening. Content reviewed by the medical team at Aquila Medical Center, Singapore. Last reviewed: 27 August 2026.
Use the score to find actionable risks
The useful question is not simply “How old am I biologically?” but “Which measurable health risks can I improve?”