← All Diagnostics

Biological age

Biological Age Testing

A guide to what aging clocks model, why their answers differ, and how to read a score without turning it into a diagnosis.

Start with the construct

“Biological age” is not one measurement

Biological-age clocks are statistical models. They combine selected biomarkers into an estimate based on patterns found in a reference dataset. Different clocks measure different signals and models, so two reports can disagree without either laboratory having made a simple arithmetic error.

The label can describe several different targets: resemblance to an average chronological age, association with later health outcomes, or an estimated pace of change. Those targets are not interchangeable, and none directly measures the total condition of a person's body.

No score is an age verdict or a validated treatment plan. It should not overrule symptoms, examination, established risk factors, or standard clinical testing.

Common approaches

Clocks can look at different layers of biology

DNA methylation clocks

These models use methylation patterns at selected DNA sites. Some were trained to predict chronological age; others were trained against health-related outcomes or change over time.

Proteomic and metabolomic models

These approaches combine circulating proteins or metabolites into an estimate. Their outputs depend on the sample, platform, reference population, and modeling choices.

Immune and glycan models

Some models use immune-linked signals such as IgG glycosylation. They may reflect a different slice of biology than methylation or multi-omic estimates.

Reading change

A trend needs a stable ruler

Sample collection, short-term illness, medication changes, laboratory processing, model updates, and ordinary biological variation can all shift a result. A small difference is not automatically evidence that aging accelerated or slowed.

Trends require comparable methods and clinical context. Before repeating a clock, decide how much change would be meaningful, whether the assay is reproducible enough to detect it, and what action would follow.

Before testing

Questions worth asking first

  1. 01What signal does this clock measure, and what outcome was it trained to estimate?
  2. 02Has the model been validated in people like me and on the same laboratory platform?
  3. 03Would a higher or lower result change a decision that is already supported by clinical evidence?
  4. 04If I repeat it, can the same sample method, assay, laboratory, and model version be used?

Continue reading

Compare this overview with the guides to multi-omic SystemAge models, IgG-glycan age estimates, and decision-led laboratory testing.