This review maps the entire toolbox for measuring “biological age” — from a stopwatch timing how fast you walk, to DNA methylation clocks, to AI that guesses your age from a selfie or an ECG — and argues that despite impressive mortality-prediction power, almost none of these tools have yet been proven to change when you intervene, which is the one thing that matters for longevity medicine.
Two people can share a birthday and yet be biologically decades apart. One climbs stairs two at a time at 70; the other is frail at 60. Chronological age — years since birth — is a blunt instrument that hides this variation. The “big idea” of this review from Northwestern’s Feinberg School of Medicine is that aging can be measured directly, and that the field now has three broad families of tools to do it.
The first family is old-fashioned and surprisingly powerful: physical function. Gait speed, grip strength, balance, and aerobic capacity each predict death and disease across hundreds of thousands of people. Gait speed has even been called a “sixth vital sign.” The second family is molecular: epigenetic “clocks” that read chemical tags on DNA. These have evolved through generations — first-generation clocks (Horvath, Hannum) simply guessed your age; second-generation clocks (PhenoAge, GrimAge) were trained on disease and death and predict mortality far better; and third-generation tools like DunedinPACE measure the speed at which you are aging right now, making them especially attractive for testing whether an intervention works. Parallel efforts read proteins, sugars (glycans), and metabolites.
The third and newest family is digital and AI-driven. Deep-learning models now estimate a person’s age — and the gap between that estimate and their real age — from retinal photographs, brain MRIs, chest X-rays, ECGs, ordinary facial photos (“FaceAge”), and even the free text of medical records via large language models. A bigger gap generally means higher mortality risk.
The honest punchline, which the authors deliver repeatedly, is that this is overwhelmingly correlation, not proven causation or modifiability. The evidence is dominated by cross-sectional and prognostic studies. There is little proof that nudging any of these clocks downward actually buys you healthier years, that short-term changes exceed measurement noise, or that different clocks even agree with one another. The field lacks a consensus standard, head-to-head comparisons, and — critically — prospective trials showing that moving the needle on a clock moves the needle on health. The authors call for harmonized, multi-site trials that treat biological age as a modifiable endpoint, while warning loudly against direct-to-consumer “reverse your age” marketing built on unvalidated tests.
Actionable Insights
This is a review of measurement tools, not an intervention, so the take-home is about what to track and how seriously to take it — plus the genuinely large effect sizes attached to a few simple, cheap markers.
The strongest, most replicated signals are functional and nearly free to measure. Grip strength: in the PURE study of 139,691 adults, each 5 kg drop in grip strength was associated with a 16% higher all-cause mortality risk — a marker you can track with a roughly $30 dynamometer. Gait speed: each 0.1 m/s faster walking carried a 10–12% lower mortality risk in a pooled analysis of 34,000+ older adults. Aerobic capacity (VO2max): declines about 6–8% per decade and is one of the strongest predictors of long-term survival — arguably the single most modifiable item on this list through exercise.
Practical message: the cheapest, best-validated “biological age” readouts you can act on today are physical — strength, walking speed, and cardiorespiratory fitness — and all three respond to training. For molecular and digital clocks (epigenetic tests, retinal-age, FaceAge), the honest take is to treat them as risk estimates with real uncertainty, not as a verdict, and not as a basis for starting or stopping therapies. A 0.5-year clock change over a few months is, per the authors, likely noise. [Confidence: High] for the functional-marker effect sizes; [Confidence: Low] for any claim that lowering a molecular/digital clock extends your life.
Source:
- Open Access Paper: Functional, molecular, and digital measurements of biological age.
- Institution: Northwestern University Feinberg School of Medicine (with co-author affiliation at Shirley Ryan AbilityLab), Chicago, Illinois.
- Country: United States.
- Journal: The Journal of Clinical Investigation (JCI).
- Impact Evaluation: The impact score of this journal is 13.6 (2024 Journal Impact Factor), evaluated against a typical high-end range of 0–60+ for top general-science and clinical journals; therefore this is a High impact journal.