What does a biological-age number actually mean?
The number means “this model mapped these inputs to this age-like scale.” Its meaning depends on what the model learned. Horvath's first-generation DNA-methylation clock was designed to estimate chronological age across tissues [2]. DNAm PhenoAge was trained through a clinical composite linked to aging outcomes [3]. DunedinPACE was designed to estimate how quickly multiple systems are changing, with 1.0 representing one biological year per chronological year in its scale [4].
These outputs can inform research and personal questions, but they do not reveal a single hidden age shared by every organ. Before interpreting a result, ask what was measured, what outcome the model predicts, who was represented in its validation cohorts, and how repeatable the test is.
How do the main biological-age methods differ?
| Method | Typical inputs | What the output is closest to | Main limitation |
|---|---|---|---|
| Chronological-age methylation clock | DNA methylation at selected CpG sites | Molecular patterns associated with calendar age | Accurate age prediction is not the same as health-risk prediction |
| Phenotypic or mortality clock | Blood chemistry, clinical data, and sometimes DNA methylation | Relative health or mortality risk on an age-like scale | Population and laboratory differences affect interpretation |
| Pace-of-aging clock | Longitudinal organ-system changes distilled into a methylation assay | Estimated rate of aging | A rate is not directly comparable with an age in years |
| Functional or fitness age | VO2 max, strength, mobility, or cardiovascular function | Performance relative to age norms | Describes selected systems, not whole-body aging |
| Wearable or questionnaire estimate | Activity, resting heart rate, sleep, fitness, and self-report | Educational risk or behavior summary | Input error and proprietary scoring can dominate the result |
The review literature treats biological age as a family of predictors, not one settled biomarker [1]. Compare methods by purpose and evidence, not by which produces the youngest-looking number.
Which biological-age estimate can you calculate yourself?
If you have appropriate laboratory results, published clinical-biomarker formulas such as Phenotypic Age can be calculated from chronological age and specified blood markers. That is different from DNAm PhenoAge, which requires methylation data even though its training target came from a clinical composite [3]. Units, transformations, missing values, and laboratory ranges matter; copying numbers into an unverified calculator can create a precise-looking error.
For a no-lab educational estimate, use the Dorsi biological age calculator. Its purpose is to show which lifestyle dimensions may be worth attention. It should not be compared directly with a methylation clock or used to diagnose disease.
Can a smartwatch estimate biological age?
A smartwatch can estimate components such as cardiorespiratory fitness, resting heart rate, activity, and sleep patterns. A model can combine them into a fitness or behavioral age, but the watch is not measuring DNA methylation, blood chemistry, or every organ system. The honest label is therefore “wearable-based estimate,” not a claim that the device found your body's true age.
Wearable inputs also have their own measurement error and missing data. Review the source metric before interpreting the composite score; Wearable Metrics Explained covers that process.
How should you track biological age over time?
Choose one method that fits the question, record the version and collection conditions, and repeat it only after enough time for a plausible change. Compare like with like: the same tissue or laboratory method, similar time of day and health state, and the same wearable window or questionnaire wording.
Look for confidence intervals, technical repeatability, and changes in the underlying inputs. A two-year improvement is not meaningful if the assay or model commonly varies by that amount. Do not average or rank age values from unrelated methods as though they share a unit.
What can you actually do about the result?
Act on validated health factors, not on the headline age. Build cardiorespiratory fitness, strength, sleep consistency, and cardiometabolic health through changes that are appropriate for your health and repeatable over months. The Strength Training guide provides a starting framework for one of those inputs.
When a score is driven by an abnormal clinical marker, follow that marker through normal medical care rather than trying to “biohack” the composite. When it is driven by self-report or wearable behavior, pick one measurable behavior and reassess the behavior first. A lower model output may be encouraging; it is not proof that aging itself was reversed.