What Is Biological Age and How Do You Measure It?

I was 41 when I first got my biological age back from an epigenetic test. The number was 38. I stared at it for longer than I expected to, not because I was relieved, but because I had no idea what to do with it. Was three years younger actually good? What was driving it? And if I changed something about my life, would the number move?

That confusion is what eventually pushed me to build Vitality AI Health with my cofounder Venkata Ramana. But before you get into our platform or our score, you should understand the underlying science. Because biological age is one of the most meaningful numbers you can track, and also one of the most misunderstood.

Chronological age versus biological age

Chronological age is simple: the number of years since you were born. What it cannot tell you is anything meaningful about your health.

Biological age is different. It is an estimate of how old your body's cells and organ systems actually are, based on measurable molecular changes. Two people can both be 45 years old chronologically and look nearly identical on the outside, while one of them shows cellular patterns consistent with a 38-year-old and the other shows patterns consistent with a 55-year-old. That gap is not cosmetic. It corresponds to real differences in mortality risk, cardiovascular function, immune resilience, and cognitive trajectory.

The core mechanism that makes biological age measurable is DNA methylation. Over the course of your life, chemical tags called methyl groups attach to specific sites on your DNA in predictable patterns. Your genome does not change, but the way it is expressed does, and that epigenetic layer accumulates information about everything your body has been through: stress, diet, sleep, inflammation, exercise, and decades of compounding choices. Researchers discovered they could read those patterns and reverse-engineer a remarkably accurate estimate of biological age.

The epigenetic clocks and what they actually measure

Not all clocks are built the same way, and each one answers a slightly different question.

Horvath clock (2013): The original pan-tissue clock, developed by UCLA's Steve Horvath using 353 CpG sites measured across 51 tissue types. It was the first clock accurate enough to predict age with a correlation coefficient near 0.96, better than almost any other biomarker at the time. The Horvath clock is good at estimating accumulated aging burden across tissues, but it was not designed to predict mortality or disease risk directly. Think of it as a baseline measure of how much epigenetic aging has happened.

Hannum clock (2013): Developed at UC San Diego using 71 markers in blood. Because it is blood-specific, it performs better in blood samples than the Horvath clock, and it tends to be more sensitive to lifestyle variables. It was an important step forward but shares the Horvath clock's limitation: predicting how old you look, more than predicting what will kill you.

PhenoAge (2018): Developed by Morgan Levine and colleagues, PhenoAge represented the first real shift toward mortality prediction. Instead of training the clock on chronological age, Levine trained it on "phenotypic age," a composite built from blood biomarkers like albumin, creatinine, CRP, white blood cell count, and several others that actually correlate with lifespan. The result is a clock that tells you something closer to biological fitness, not just biological accumulation.

GrimAge (2019): This is currently the most clinically validated clock for predicting mortality and disease risk. Developed by Ake Lu and Steve Horvath at UCLA, GrimAge takes a further step: it builds on DNA methylation surrogates of specific plasma proteins associated with mortality, including PAI-1 (a cardiovascular risk marker) and a methylation-based estimator of smoking pack-years. When Lu et al. published GrimAge in Aging in January 2019, they demonstrated it predicted time-to-death 18% more accurately than chronological age and outperformed all prior epigenetic clocks on cardiovascular disease prediction. If you could only track one clock right now, GrimAge has the strongest evidence base.

DunedinPACE (2022): This is conceptually distinct from the others. Rather than measuring biological age as a static number, DunedinPACE measures the pace at which you are currently aging: how many biological years you are accumulating per chronological year. It was developed using longitudinal data from the Dunedin birth cohort tracked across four time points over two decades, modeling decline in 19 organ-system integrity measures simultaneously. A DunedinPACE score of 1.0 means you are aging at the normal expected rate. Below 1.0 means slower. Above 1.0 means faster. Published in eLife in 2022, DunedinPACE is particularly useful for tracking the impact of interventions because it measures current rate, not accumulated burden. If you change your lifestyle and retest in six months, DunedinPACE is more likely to move than the static clocks.

What the research actually shows is possible

The most important question for anyone getting their biological age tested is: can it change? The answer is yes, though the magnitude varies depending on the clock and the intervention.

The CALERIE trial randomized 220 non-obese adults to 25% caloric restriction or normal eating for two years. A 2023 analysis published in Nature Aging found that caloric restriction slowed the DunedinPACE by 2-3%. That sounds modest, but the researchers noted that a 3% slower pace of aging corresponds to roughly a 15% reduction in mortality risk, comparable in magnitude to the effect of quitting smoking. The trial was the first randomized controlled trial to demonstrate that a behavioral intervention could move a validated epigenetic aging marker.

The TRIIM trial took a more aggressive approach. Researcher Greg Fahy and colleagues at UCLA used a combination of growth hormone, DHEA, and metformin to try to regenerate the thymus, the organ responsible for producing T-cells, which undergoes substantial atrophy after puberty. The 2019 paper in Aging Cell documented what they found: across all four epigenetic clocks they measured, participants showed a mean reversal of approximately 2.5 years of biological age over 12 months, with the effect accelerating toward the end of the treatment period. The GrimAge clock specifically showed a 2.1-year improvement that persisted six months after treatment ended. It was the first human evidence that biological age could actually be reversed, not just slowed.

For cardiovascular risk specifically, the link between biological age and outcomes is independent of traditional risk factors. A 2025 study of 1,264 MESA participants found that higher GrimAge was associated with significantly elevated risk of composite cardiovascular disease and stroke even after adjusting for demographics, behavioral factors, comorbidities, and medications. People who are biologically older than their chronological age are not just at higher risk in theory. They are experiencing measurably more cardiovascular events in the real world.

The consumer testing landscape

If you want to measure your own biological age, there are now several at-home options. Here is how the main ones compare:

Test

Methodology

Clocks Reported

Price Range

Notes

TruDiagnostic TruAge COMPLETE

Illumina EPIC array, fingerstick blood spot

OMICmAge, DunedinPACE, SYMPHONYAge (11 organ systems), PhenoAge

~$429-499

Widest consumer coverage; includes organ-system-level aging via Yale's SYMPHONYAge algorithm; exclusive DunedinPACE reporting

Elysium Index

DNA methylation from saliva; proprietary algorithm

Single biological age score

~$299/kit

Developed in partnership with Morgan Levine's group; uses blood-methylation methodology; less granular than TruDiagnostic

Thorne Biological Age

Blood draw at lab; Klemera-Doubal statistical algorithm

Biological age + aging pace vs. chronological

~$250-299

Transparent methodology (Klemera-Doubal is well-validated); single-score output limits actionability; strong brand credibility

Hannah Biosciences GrimAge

Illumina array, venous blood

GrimAge specifically

~$325-425 depending on collection method

Reports the single most mortality-predictive clock; useful if GrimAge specifically is your focus; less organ-system granularity

A few things worth knowing before you spend the money. First, no consumer test is a medical diagnostic. Second, test-retest reliability matters. DunedinPACE was specifically engineered for high reliability, which is why it moves consistently when you change behaviors. Third, a single data point is not very useful. The value comes from serial measurements spaced six to twelve months apart so you can track direction of change.

My own data, and what I did with it

When I got that first result at 41, biological age of 38, I did not assume it was a permanent advantage. I treated it as a baseline. At the time I was training moderately, sleeping seven hours most nights, and eating without much structure. My DunedinPACE was 0.91, meaning I was aging about 9% slower than my chronological peers.

I spent the following six months prioritizing a few specific changes: consistent strength training four days per week, two additional sleep quality interventions (cooler room, no alcohol within three hours of bed), and a modest caloric reduction that was not tracked obsessively but was deliberate. Six months later, DunedinPACE had moved to 0.87. Biological age was 37.5 against a chronological age of 41.5.

That is a small number. But across a lifetime, small differences in pace compound significantly. The goal was never a flattering number. The goal was to understand which inputs were actually moving it.

The interventions that have real evidence behind them

Caloric restriction has the strongest randomized trial evidence, courtesy of CALERIE. The finding is not that you need to be starving. CALERIE participants achieved an average of only 11.7% restriction, well below the 25% target, and still showed measurable slowing of biological aging.

Exercise matters, with increasing specificity emerging in the research. Resistance training appears to be particularly protective for muscle-system biological age and metabolic markers. Zone 2 aerobic training (low-intensity, conversational pace) is consistently associated with slower cardiovascular and metabolic aging in observational data. The combination appears to be more powerful than either alone.

Sleep is possibly underrated as a biological age lever. Deep sleep is when cellular repair mechanisms are most active, and chronic sleep restriction is associated with accelerated GrimAge in multiple observational studies. Getting from six hours to seven-and-a-half hours of quality sleep may do more for your DunedinPACE than adding a supplement.

On supplementation: the evidence base is thinner and more mixed. Resveratrol, NAD+ precursors, and rapamycin all have theoretical mechanisms and animal data, but rigorous human randomized controlled trial evidence specifically showing biological age movement is limited. That picture is changing quickly, but anyone selling you supplements primarily on the basis of epigenetic clock data is getting ahead of the science.

How Vitality AI Health uses this

At Vitality AI Health, we built our VAI Score to aggregate biological age signals alongside other data streams (blood work, DEXA body composition scans, wearable data, and your own reported inputs) into something that tells you not just what your age is, but what the primary drivers of acceleration or slowdown appear to be for your specific biology.

The epigenetic clock is one critical input, but it is not the only one. Your fasting glucose trend, your resting heart rate variability, your visceral fat from DEXA: these all contribute to the picture. What our platform tries to do is make sense of those signals together, through Vaya Chat, our AI assistant that helps you interpret what your data actually means and what would be worth changing.

If you want to understand why I built this, and what problem I was actually trying to solve, that story lives at /blog/why-i-built-vitality-ai-health.

Where this is going

The clocks are getting better fast. Second-generation clocks like GrimAge already add independent predictive information beyond traditional cardiovascular risk scores. The researchers working on third-generation clocks are training on multi-omic data (proteomics, metabolomics, and transcriptomics combined) that will likely push predictive accuracy much higher.

What that means practically is that within a few years, the difference between being biologically 38 and biologically 45 at a chronological age of 41 will be something physicians use in clinical decisions, not just something health-obsessed founders track on spreadsheets. The infrastructure for acting on that data (personalized interventions, longitudinal tracking, AI-assisted interpretation) needs to exist before the tests become standard of care. That is the gap we are building toward.

Your biological age is not fixed, and the research is clear on that point. What is less clear, for most people, is how to interpret the number they get back and what to change first. That is exactly what we are working on.

Citations and sources: Lu et al., GrimAge, Aging 2019 | Fahy et al., TRIIM trial, Aging Cell 2019 | Belsky et al., CALERIE DunedinPACE, Nature Aging 2023 | Epigenetic Aging Clocks and Cardiovascular Outcomes, JAHA 2025 | Belsky et al., DunedinPACE, eLife 2022