A biological-age clock built from ordinary health records sounds less glamorous than one built from DNA methylation or advanced imaging. That may be precisely why it matters. Routine clinical records already contain years of blood tests, diagnoses and measurements for millions of people. A team reporting in Nature Medicine used that accumulated information to ask whether the pace of development and aging can be estimated from the medical data health systems collect every day.
Evidence at a glance
- Study type: Large retrospective modeling study with external validation.
- Population: More than 24.6 million clinical visits from about 9.7 million people.
- Evidence level: Strong observational and predictive evidence, not proof of a biological mechanism.
- Main caution: A clock trained on health records can partly measure illness, treatment and access to care rather than aging itself.
What the researchers built
The system, called LifeClock, was developed from longitudinal records spanning childhood through later adulthood. The researchers selected 184 commonly available clinical indicators, including laboratory values and other features recorded during routine care. They then built separate models for pediatric development and adult aging. That distinction is important because a child moving toward physiological maturity is not undergoing the same process as an adult accumulating age-related risk.
In adults, several influential signals were familiar markers of health and physiological reserve. Higher urea, lower albumin and higher red-cell distribution width contributed substantially to the model. None is an aging switch on its own. Together with many other measurements, however, they can describe patterns that become more common as organ function, inflammation, nutrition and disease burden change with age.
What the clock could predict
The model estimated chronological age with useful accuracy and also carried information about current and future disease. In an external test using UK Biobank data, the adult clock produced a mean absolute error of about 4.14 years. People whose clinical profile appeared older than their calendar age tended to carry greater health risk.
That makes the approach potentially useful for research and population health. A health system might identify groups whose physiology is deteriorating faster than expected, while clinical researchers could use the measure to stratify participants or monitor broad changes. Because the inputs are already collected, the method may be easier to deploy than a specialized assay.
The central limitation
Prediction is not the same as measurement of a root cause. A person with chronic disease will often have abnormal laboratory values, more medical visits and more treatment. A record-based clock may therefore recognize the consequences of disease and the structure of healthcare use. It does not automatically reveal a distinct biological aging process underneath them.
There is also a risk of data leakage and population bias. Clinical records are generated for care, not for a perfectly standardized experiment. Tests are ordered for reasons, reference ranges vary, missing data are meaningful and access to care is unequal. Performance in one health system does not guarantee equivalent performance across countries, ancestries or socioeconomic groups.
What would increase confidence
The strongest next step would be prospective validation. Researchers should test whether changes in LifeClock reliably precede meaningful outcomes, whether the score responds appropriately to proven risk-reducing interventions and whether it adds information beyond conventional clinical risk models. Calibration across diverse health systems will also matter.
The Lifespan Brief assessment
LifeClock is a promising infrastructure tool, not a verdict on an individual’s biological age. Its scale and use of ordinary records are genuine strengths. Its value will depend on whether it predicts outcomes fairly and prospectively, and whether clinicians can act on the information. For now, it is best understood as a sophisticated risk signal built from the medical footprints that aging and disease leave behind.