Contents

1. What is biological age?
2. Overview of concepts underpinning epigenetic clocks
   1. Epigenetics
   2. DNA methylation
3. Why aging leaves methylation marks
4. Biological aging clocks
   1. The original Horvath clock
   2. The Hannum clock
   3. PhenoAge
   4. GrimAge
   5. DunedinPACE
   6. Which clock is best?
   7. What clocks capture — and what they miss
      1. Senescent cells and senolytic effects may not be captured well.
      2. Telomere length is only weakly related to many methylation clocks.
      3. Somatic mutations are not directly measured by methylation.
      4. Blood-based clocks can miss tissue-specific aging.
5. Interventions that affect epigenetic aging
   1. Calorie restriction and weight loss
   2. Omega-3, vitamin D, and exercise
   3. Correcting vitamin D deficiency
   4. Multivitamins
   5. Dietary patterns, meat, vegetables, and carotenoids
      1. Does red meat accelerate epigenetic aging?
   6. Exercise and physical activity
   7. Deliberate exercise may be a stronger epigenetic lever
   8. Body temperature, metabolism, and hibernation
   9. Can sauna mimic exercise-like signals?
   10. Sleep disruption
   11. Social connection and loneliness as age accelerators
   12. Medical interventions
       1. HIV and antiretroviral therapy
       2. Anti-TNF therapy
       3. Metformin
       4. GLP-1 therapy and semaglutide
6. Can AI build better aging clocks?
7. Consumer biological age tests
   1. Consumer Test List
      1. What to Look For Before Buying
      2. Best Broad Option
      3. More Specific Options
      4. Simple Consumer Scores
      5. Disclaimer: Why two epigenetic age tests might disagree
8. Conclusion

Biological aging clocks are among the most important tools in modern longevity science. They attempt to answer a deceptively simple question: can we measure how fast a person is aging biologically, rather than just counting the number of years they have been alive?

The answer is yes—but with important caveats.

Epigenetic clocks use patterns of DNA methylation, a chemical tagging system on DNA, to estimate different aspects of aging. Some clocks estimate chronological age. Others predict disease risk, mortality risk, or the current pace of biological aging. This means biological age is not one simple number. Different clocks are built for different purposes, and disagreements between clocks can be biologically informative rather than merely confusing.

This topic page explains what biological age is, how DNA methylation clocks work, what the major clocks measure, where they perform best, what they miss, and which interventions currently have the strongest human evidence for moving them.

What is biological age?

"Biological age is not one thing. It can mean functional age, molecular age, organ age, mortality-risk age, or pace of aging." Click To Tweet

Chronological age is simple: it is the number of years since birth.

Biological age is more complicated. It reflects the fact that two people of the same chronological age can differ dramatically in disease risk, physical function, immune health, metabolic health, organ function, and mortality risk.

Geroscientists—researchers who study aging—operationalize biological age through measurement technologies: molecular markers, blood biochemistry, organ-function measures, VO2 max, gait speed, frailty, and other functional readouts.

Biological age can also be estimated from a distinct molecular process known as methylation—a chemical tagging system on DNA that helps regulate which genes are turned on or off.

Overview of concepts underpinning epigenetic clocks

To understand why DNA methylation can be used as a biological clock, it helps to first understand what epigenetics is and how methylation regulates gene activity.

Epigenetics

Epigenetics is a biological mechanism that regulates gene expression (how and when certain genes turn on or off). Diet, lifestyle, and environmental exposures can drive epigenetic changes throughout an individual's lifespan to influence health and disease. For example, epigenetic processes dysregulate in diseases such as cancer and Alzheimer's disease.^[1][2] Scientific evidence suggests that epigenetic changes can be passed from generation to generation.^[3]

Three biochemical processes drive epigenetic change: DNA methylation, histone modification, and non-coding RNA-associated gene silencing. DNA methylation has relevance for predicting biological age via epigenetic clocks.

DNA methylation

DNA methylation occurs when a methyl group – a chemical structure containing three hydrogen atoms and one carbon atom – attaches to one of DNA's four nucleotide bases (adenine [A], cytosine [C], guanine [G], or thymine [T]). A class of enzymes called methyltransferases facilitates the process of DNA methylation, while another type, the ten-eleven transferases, or TET enzymes, reverses it.^[4] Methylation creates a biological record of the varied molecular processes that participate in an individual's development, maintenance, and decline. Many factors promote methylation, including dietary intake, exercise, stress, smoking, and even social factors, such as maternal-infant interaction.^[5]

The most common DNA methylation process involves adding a methyl group to one of the carbon atoms in the cytosine base, forming 5-methylcytosine. This addition alters the overall geometry of the DNA strand, ultimately influencing gene expression. Most 5-methylcytosine is on areas of the DNA known as CpG islands – short stretches of DNA where the frequency of the cytosine-guanine (CG) sequence is higher than in other regions. (The "p" in CpG reflects the presence of a phosphate group between the two nucleotides.) Methylation of a CpG island in the promoter region of a gene turns off, or "silences," the gene's expression. DNA methylation may promote age-related diseases such as cancer.^[6]

Methylation is a dynamic process that is reversible and appears to be under circadian control in some tissues.^[6][7] It increases with age, and the rate of change over the lifespan varies among CpGs, averaging 3.2 percent organism-wide and ranging from 7 to 91 percent for specific individual genes.^[8][9]

Why aging leaves methylation marks

Aging can be described as the accumulation of damage and dysfunction across biological systems: proteins, metabolites, immune signaling, mitochondria, stem cells, tissue structure, and the regulatory environment surrounding DNA. DNA methylation clocks may capture a form of damage accumulation that occurs "deep inside" cells even in people who do many things right.

DNA contains the instructions, but cells must know which instructions to use. A liver cell and a neuron contain essentially the same DNA, but they behave differently because different genes are active or silent. DNA methylation helps regulate that cellular identity.

With aging, methylation marks can be gained or lost in the wrong places. One way to describe this is a "flattening of the methylation landscape": regions that should be sharply methylated or unmethylated become less distinct. That may impair cell identity and cellular function over time.

Methylation may also act as a biological memory of long-term exposures. Smoking, inflammation, obesity, metabolic dysfunction, infection, psychological stress, air pollution, and other stressors may leave durable regulatory marks. Those marks may not be the only cause of aging, but they can reveal the body's cumulative exposure to aging-related insults. One example of this comes from research into the long-term consequences of prenatal exposure to the Dutch famine of 1944–45, including a historical cohort of 2,414 children born in Amsterdam around the 1944–1945 famine. Prenatal famine exposure was defined as average maternal rations below 1,000 calories per day during any 13-week period of gestation.^[10]

Early gestation exposure to famine often produced the clearest adult-life effects even without reduced birth size. Reported outcomes include:

• altered glucose tolerance

• higher type 2 diabetes risk

• more atherogenic lipid profiles

• increased coronary heart disease and earlier onset

• altered food preference

• mental-health differences

• lower physical function in men

• cognitive and brain-volume differences

• evidence of DNA methylation differences

Late-gestation exposure was more directly linked to smaller body size at birth and later glucose abnormalities.

This is a vivid example of the "memory of stressors" concept: a time-limited exposure can leave biological signatures that shape disease risk decades later.

Biological aging clocks

"The dream of the field is a validated surrogate endpoint: a biomarker that changes after an intervention and reliably predicts better long-term health or survival. We are not fully there yet." Click To Tweet

A major goal of biological aging research is to develop tools that can test whether interventions actually slow or reverse aspects of aging. This is critical because waiting decades for mortality outcomes is impractical. A useful aging biomarker could help identify promising interventions earlier, while still requiring long-term validation.

The dream of the field is a validated surrogate endpoint: a biomarker that changes after an intervention and reliably predicts better long-term health or survival.

Epigenetic clocks are among the most promising candidates, but they are not there yet. Clock changes can provide useful evidence that an intervention has altered aging-related biology, but they should not be treated as proof that a person has gained a specific number of years of life.

The original Horvath clock

"The original Horvath clock made biological aging measurable at the molecular level, but it was trained primarily as a chronological-age estimator." Click To Tweet

The original Horvath clock is the foundational epigenetic aging clock.

It was trained to estimate chronological age from DNA methylation across a broad range of human tissues and cell types. Steve Horvath developed a multi-tissue DNA methylation age predictor using approximately 8,000 samples from 82 methylation datasets spanning 51 healthy tissues and cell types.^[8]

• DNAm age was near zero in embryonic and induced pluripotent stem cells, increased with cell passage number, and worked across a broad range of tissues.

• In cancer, all 20 cancer types analyzed showed substantial age acceleration, averaging about 36 years.

• The major innovation of the Horvath clock was not that it was the best mortality predictor, but that it worked across many tissues, making it useful for asking whether a tissue, cell type, tumor, or reprogrammed cell has an unusually old or young methylation profile.

DNA methylation age may reflect the cumulative effect of an "epigenetic maintenance system," and the clock has been used in developmental biology, cancer biology, tissue-aging comparisons, and cellular reprogramming studies.

The best evidence that the original Horvath clock captures something biologically meaningful beyond calendar age is that age acceleration predicts mortality, but its mortality signal is modest compared with later clocks. In a 2015 meta-analysis of four cohorts, a 5-year higher Horvath DNAm age acceleration was associated with a 9% higher mortality risk, whereas the blood-based Hannum clock was somewhat stronger.^[11]

This is why the original Horvath clock is best framed as a cross-tissue biological age benchmark, not the preferred clock for mortality prediction.

The Hannum clock

"The Hannum clock is best understood as an adult blood-based epigenetic age clock: highly useful for studying blood and immune-system aging, but not a universal tissue clock." Click To Tweet

The Hannum clock was one of the first major DNA methylation aging clocks and remains an important reference point in the history of epigenetic aging research. Developed by Hannum and colleagues, it estimates epigenetic age from DNA methylation patterns at 71 CpG sites in human blood.^[12] Like the original Horvath clock, the Hannum clock is considered a first-generation epigenetic clock because it was trained primarily to predict chronological age rather than disease, mortality, or pace of aging. But unlike the original Horvath clock, which was designed to work across many tissues, the Hannum clock was developed from adult whole-blood samples. That makes it especially relevant for studying aging signals in blood, immune cells, and circulating leukocyte composition, but less appropriate for estimating the age of other tissues or for use in children.

The original study analyzed genome-wide DNA methylation profiles from human blood and identified a 71-CpG model that accurately estimated age across adulthood. Many of the age-associated methylation changes occurred near genes involved in development, metabolism, immune function, and cellular regulation, supporting the idea that aging leaves a coordinated methylation signature in blood.^[12]

The Hannum clock also showed early evidence that DNA methylation age acceleration was biologically meaningful. In a later meta-analysis of four cohorts, both the Horvath and Hannum clocks predicted all-cause mortality independent of chronological age and other risk factors. In that analysis, a 5-year higher Hannum DNA methylation age acceleration was associated with a 21% higher mortality risk, whereas the corresponding estimate for the Horvath clock was 9%.^[11]

The Hannum clock has been used to study early-life adversity. In one study, exposure to abuse, financial hardship, or neighborhood disadvantage around age 7.5 years was associated with altered DNA methylation patterns, including methylation signatures related to cellular aging.^[13]

This fits a broader theme in epigenetic aging research: the methylome can act as a biological memory system, recording signals from stress, development, inflammation, and environment long before clinical disease appears.

The best use of the Hannum clock is therefore not as a universal biological-age score, nor as a modern mortality-risk predictor. It is most useful as a blood-based, first-generation methylation age measure that helped establish the field and remains valuable for studies of immune aging, social exposures, environmental stress, and adult blood methylation biology. Its main limitation is also its strength: because it was built in adult blood, it can be informative about blood aging but should not be assumed to represent every tissue, every developmental stage, or every dimension of biological aging.

PhenoAge

"PhenoAge is a bridge between routine clinical biology and methylation clocks. It asks whether the methylome looks like the methylome of someone with a higher-risk clinical aging phenotype." Click To Tweet

PhenoAge was designed to move beyond "how old does the methylation pattern look?" toward "how old does the person's physiology look?"

It was developed by Morgan Levine and colleagues using NHANES data with long-term mortality follow-up. The goal was to create a methylation-based measure that captured a clinical aging phenotype rather than chronological age alone. Levine and colleagues selected nine blood biomarkers plus chronological age from 42 clinical markers: albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red cell distribution width, alkaline phosphatase, white blood cell count, and age. They then trained a DNA methylation predictor of that clinical phenotypic age using 513 CpGs.^[14]

DNAm PhenoAge outperformed earlier chronological-age clocks for predicting several aging outcomes, including all-cause mortality, cancers, healthspan, physical functioning, and Alzheimer's disease-related outcomes.

• Each 1-year higher DNAm PhenoAge was associated with a 4.5% higher risk of all-cause mortality.

• The top 5% "fastest agers" had a mortality hazard about 1.62 times that of the average person and 2.58 times that of the slowest agers.

• The clock also predicted comorbidity burden, disease-free status, physical functioning, and coronary heart disease risk.

GrimAge

"GrimAge is a hazard signal, not a horoscope. It estimates relative mortality risk; it does not tell you when you will die." Click To Tweet

GrimAge is one of the most important clocks to discuss carefully because its name and output can be misleading. It is not a countdown clock. It is a mortality-risk model expressed in age-like units.

GrimAge was built in two stages. First, the authors developed DNAm surrogates for plasma proteins and smoking pack-years. Then they combined selected DNAm protein surrogates, DNAm smoking pack-years, chronological age, and sex into a predictor of time-to-death.

The main development and testing used the Framingham Heart Study Offspring cohort, with additional validation in several large cohorts. The final GrimAge model selected DNAm surrogates for seven proteins—adrenomedullin, beta-2-microglobulin, cystatin C, GDF15, leptin, PAI-1, and TIMP1—plus DNAm smoking pack-years, age, and sex. The final score involved 1,030 unique CpGs.^[15]

• AgeAccelGrim showed strong associations with time-to-death, coronary heart disease, cancer, fatty liver, visceral fat, age at menopause, and comorbidity count.

• DNAm PAI-1 was especially strongly linked to lifespan, type 2 diabetes, and comorbidity.

In two separate cohorts, a 1-standard deviation higher GrimAge acceleration was associated with a roughly 2-fold higher risk of all-cause death, a 44% higher risk of myocardial infarction, and a 42% higher risk of stroke. GrimAge also outperformed Horvath, Hannum, and PhenoAge across clinical aging phenotypes and remained associated with walking speed, polypharmacy, frailty, and mortality.^[16]

GrimAge is best understood as a measure of mortality hazard compared with people of the same chronological age and sex. If a 50-year-old has a GrimAge of 58, that means their near-term mortality risk is higher than expected for a typical 50-year-old of the same sex. It does not mean they will die at 58.

GrimAge is one of the strongest clocks for mortality risk and disease-risk prediction, especially when the question is not "what is the pace of aging?" but "how much long-term risk is encoded in this blood methylation profile?" Its strongest signals likely come from smoking exposure, inflammatory/proteomic stress, cardiometabolic risk, kidney-related risk, immune aging, and other long-term exposures that leave methylation marks.

DunedinPACE

"DunedinPACE is best described as an odometer or speedometer: it estimates the current pace of multisystem biological decline, not how many years old the body looks." Click To Tweet

DunedinPACE was designed to measure the speed of aging rather than accumulated age. It is best understood as an "odometer" or "speedometer" for the pace of biological aging, in contrast to clocks that estimate accumulated biological age or mortality risk.

DunedinPACE was developed using the Dunedin cohort, a birth cohort in New Zealand, by measuring changes in 19 indicators of organ-system integrity across ages 26, 32, 38, and 45 years. These repeated measures covered cardiovascular, metabolic, renal, hepatic, immune, periodontal, and pulmonary systems. The resulting longitudinal pace-of-aging phenotype was then used to train a single-timepoint blood DNA methylation algorithm. DunedinPACE is scaled so that a value of 1.0 represents one year of biological change per calendar year.^[17]

• The slowest-aging participants were around 0.40 biological years per calendar year and the fastest were around 2.44.

• The biomarker showed high test-retest reliability, was associated with morbidity, disability, physical functioning, cognitive decline, and mortality risk, and added information beyond GrimAge in some validation analyses.

• Pace of aging was also faster among young people exposed to childhood adversity and poverty, a known aging accelerator.

DunedinPACE is therefore well suited for intervention trials and exposure studies where the question is whether an intervention changes the rate at which aging-related physiology is moving. It has unusually high test-retest reliability for an epigenetic aging measure, and in validation work it was associated with morbidity, disability, and mortality while adding predictive information beyond GrimAge.

Which clock is best?

"For disease prediction, use the second- and third-generation clocks. For mechanistic aging biology, first-generation clocks can still be informative, but they are weaker clinical predictors." Click To Tweet

The best clock depends on the question.

• For chronological-age estimation, the original Horvath clock is foundational.

• For mortality and disease prediction, second-generation clocks such as GrimAge and GrimAge2 often perform better.

• For intervention studies, DunedinPACE may be especially useful because it was designed to estimate rate of aging rather than accumulated biological age.

A 2025 study provided an unbiased comparison of 14 epigenetic clocks in relation to 10-year onset of 174 disease outcomes in 18,859 individuals.^[18] Second-generation clocks significantly outperformed first-generation clocks. The study found 176 significant clock-disease associations, and adding a clock to traditional risk-factor models improved classification accuracy by more than 1%. Only 9 significant disease associations came from first-generation clocks, about 5% of all significant findings. GrimAge v2 had the largest and most significant association with 10-year all-cause mortality, while the broader disease analysis suggested there is no single best clock for every disease category. Disease-specific examples included primary lung cancer with GrimAge v1, cirrhosis with GrimAge v2, and diabetes with DunedinPACE.

What clocks capture — and what they miss

Clocks should be framed as integrators. They are not measuring one hallmark of aging. They appear to integrate signals related to inflammation, smoking exposure, metabolic dysfunction, immune-cell composition, stem-cell function, plasma-protein signatures, and organ stress.

Methylation clocks do not capture all forms of aging damage equally well. They may not capture senescent cells, radiation damage, and telomere attrition, for example. In other words, a clock can move in a favorable direction while other forms of damage remain.

Senescent cells and senolytic effects may not be captured well.

Senescent cells are cells that have stopped dividing but don't die when they normally should—they accumulate with aging, tissue damage, obesity, chronic disease, or stress.

Epigenetic clocks may not fully capture senescent-cell burden or the effects of senolytics (compounds or therapies designed to selectively clear senescent cells). Senescence is a distinct aging process, and at least some forms of cellular senescence appear to be partially separable from epigenetic-clock aging.^[19]

Telomere length is only weakly related to many methylation clocks.

Telomere length and DNA methylation clocks are related to aging but are not interchangeable. They appear to capture partly distinct biology. One study noted that although DNA methylation (DNAm) age and telomere length are both associated with chronological age, they are only weakly correlated with each other, suggesting distinct underlying mechanisms governing each process. In meta-analyses, telomere length correlates only modestly with chronological age.^[20]

Somatic mutations are not directly measured by methylation.

Standard DNA methylation clocks do not directly measure somatic mutation burden—a separate form of molecular damage that requires mutation-detection approaches such as whole-genome sequencing. They measure methylation states at CpG sites.^[8]

Methylation clocks can integrate signals related to systemic aging, but physiological aging—tissue architecture, fibrosis, vascular stiffness, muscle performance, mitochondrial capacity, and brain function often require direct imaging, histological, or functional measurements to detect changes due to interventions. This is why certain clocks (i.e., DunedinPACE) may be useful—they were trained against long-term changes in indicators of organ-system integrity across cardiovascular, metabolic, kidney, liver, immune, dental, and pulmonary systems.^[17]

Blood-based clocks can miss tissue-specific aging.

Blood-based clocks are convenient and often well validated, but blood is not every organ. Tissue-specific aging can diverge from blood-based estimates, especially in organs and tissues with distinct cell composition, turnover, injury, fibrosis, or repair dynamics. In fact, there is substantial evidence that different organs and tissues age at distinct rates.^[8]

Interventions that affect epigenetic aging

"Big reversals usually start from bad baselines. The more accelerated someone is at baseline, the more room there is to improve." Click To Tweet

One of the most important principles in interpreting epigenetic aging intervention studies is that clock changes are usually largest when people begin with accelerated aging, obesity, inflammation, metabolic dysfunction, nutrient deficiency, infection, or another correctable biological stressor.

"Reversing biological age" is often presented as though healthy people can easily erase years from their aging clocks. In reality, large clock shifts often reflect the removal of a biological accelerator. Treating HIV, reducing chronic inflammation, losing substantial excess fat, correcting vitamin D deficiency, or improving cardiometabolic dysfunction can move clocks more dramatically than adding a modest supplement to an already healthy baseline.

Calorie restriction and weight loss

"CALERIE is strong human evidence that caloric restriction can slow DunedinPACE by about 2-3%." Click To Tweet

Excess adiposity is not inert storage. It is metabolically active tissue that can amplify inflammation, insulin resistance, glucose dysregulation, immune stress, and adipokine signaling—aging-relevant signals that major epigenetic clocks can detect.

But obesity does not necessarily accelerate every tissue equally. The liver appears especially sensitive. One study reported an unexpectedly strong correlation between high body mass index and the epigenetic age of liver tissue, potentially helping explain why obesity is associated with earlier onset of liver disease and liver cancer.^[21]

That is why weight loss can show up clearly in aging-clock studies, especially when the starting point is obesity or metabolic dysfunction. DunedinPACE is particularly relevant because it was trained on longitudinal changes in physiological systems that included BMI, waist-to-hip ratio, glucose impairment, and inflammation. When weight loss is modest, only the most metabolically sensitive clocks may move. But when fat loss is large enough, the signal may become broad enough for multiple clocks to detect.

One of the most important randomized human trials in this area is CALERIE Phase 2, a multi-center US randomized controlled trial in healthy adults without obesity. Participants were randomized 2:1 to a behavioral intervention prescribing 25% caloric restriction or to an ad libitum control for two years. Blood DNA methylation was measured at baseline and follow-up in 197 participants, and the primary clock analysis focused on PC-PhenoAge, PC-GrimAge, and DunedinPACE.^[22]

• The achieved caloric restriction was much lower than prescribed, averaging 11.9%.
• Even so, caloric restriction reduced DunedinPACE at 12 months and maintained the reduction at 24 months, a change corresponding to roughly 2–3% slower pace of aging.
• PhenoAge and GrimAge did not significantly change.
• Participants who achieved more than 10% caloric restriction had larger DunedinPACE effects than those who achieved less than 10%.

This pattern is important: calorie restriction appeared to slow a pace-of-aging measure without clearly changing mortality-risk clocks over the two-year period. That does not mean calorie restriction "failed." It means DunedinPACE may be more sensitive to short-term physiological slowing, while GrimAge and PhenoAge may require larger, longer, or more disease-risk-specific changes to move.

Omega-3, vitamin D, and exercise

"DO-HEALTH supports a modest omega-3 effect on multiple epigenetic clocks, with possible additive benefit when combined with vitamin D and exercise." Click To Tweet

The DO-HEALTH Bio-Age analysis is a good example of how to interpret modern biological aging studies without hype. The effect size was small—measured in months over three years—but the trial was randomized, the intervention was practical, and the signal appeared across multiple modern clocks.^[23]

The analysis included 777 older Swiss adults with DNA methylation data at baseline and three years. Participants were assigned in a factorial design to vitamin D at 2,000 IU/day, omega-3 at 1 g/day, a simple home exercise program, combinations of these interventions, or placebo/control conditions.

• Omega-3 alone slowed PhenoAge, GrimAge2, and DunedinPACE.
• Vitamin D alone and the home exercise program alone were not associated with changes in the clocks.
• The combination of all three interventions showed additive benefit on PhenoAge.
• The authors translated the standardized effects into approximately 2.9–3.8 months of biological aging difference over three years.

The fact that 88% of participants were already physically active matters. A simple home exercise program may not be enough of a stimulus to move blood methylation clocks in already active older adults. Similarly, vitamin D may not show an independent effect if many participants are not deficient. The most defensible interpretation is that omega-3 produced a modest but measurable signal across several modern clocks, while vitamin D and exercise effects may depend more strongly on baseline status and intervention intensity.

Correcting vitamin D deficiency

"Vitamin D is not an anti-aging shortcut. It is a deficiency-correction story. The clock signal is most likely to appear when supplementation fixes a real biological gap." Click To Tweet

The DO-HEALTH trial found no independent effect of vitamin D supplementation on epigenetic clocks,^[23] but this should not be interpreted as "vitamin D does nothing." The more precise interpretation is that vitamin D supplementation may not move biological-aging clocks when people are already vitamin D sufficient or when the dose comparison is modest.

The deficiency-correction model is supported by longitudinal data from the Berlin Aging Study II and GendAge follow-up. In this study, older adults were reexamined an average of 7.4 years later, and vitamin D-deficient participants who started supplementation after baseline were compared with untreated deficient participants and sufficient controls.^[24]

• Vitamin D-deficient participants who started supplementation had 2.6-year lower age acceleration and 1.3-year lower Horvath DNAm age acceleration compared with participants who remained deficient and untreated.
• DNAm age acceleration did not significantly differ between successfully treated deficient participants and healthy controls, and the effects were not consistently detected across all clocks.

This was not a randomized trial, so confounding is possible. But it supports a practical interpretation that vitamin D deficiency may be an age-acceleration state, and correcting deficiency may reduce that specific signal.

Multivitamins

"A daily multivitamin may modestly slow certain second-generation epigenetic aging clocks, particularly in people with accelerated biological aging or subtle nutritional inadequacies. Think"nutritional insurance with measurable molecular signals,"not anti-aging pill." Click To Tweet

One of the more interesting findings in the epigenetic aging literature comes from COSMOS, the COcoa Supplement and Multivitamin Outcomes Study. In this trial, a multivitamin-multimineral supplement produced a small but measurable slowing of second-generation aging clocks in older adults, especially GrimAge and PhenoAge.^[25]

The epigenetic aging analysis included 958 COSMOS participants free of major morbidity during the two-year blood-collection window. Participants received daily Centrum Silver multivitamin-multimineral and/or cocoa extract supplying 500 mg cocoa flavanols per day, including 80 mg epicatechin.

• Compared with placebo, the multivitamin slowed the yearly increase in PCGrimAge by -0.113 years per year and PCPhenoAge by -0.214 years per year.
• At two years, the difference between multivitamin and placebo was -0.209 years for PCGrimAge and -0.443 years for PCPhenoAge. First-generation clocks were not significantly affected.
• DunedinPACE increased in the placebo group and stayed stable in the multivitamin group, and the multivitamin had stronger PCGrimAge effects among people with accelerated biological aging at baseline.
• Cocoa extract did not affect any of the five clocks.

The most cautious interpretation is that multivitamins may produce small, detectable clock effects in older adults, especially when baseline nutritional gaps or accelerated biological aging are present. That is different from saying a multivitamin "reverses aging." A better framing is nutritional sufficiency with measurable molecular effects.

The COSMOS cognition data make the story more compelling. In COSMOS-Clinic and a meta-analysis of three COSMOS cognitive substudies, daily multivitamin supplementation improved global cognition and episodic memory, with the global cognition effect estimated as equivalent to roughly two years less cognitive aging.^[26] On the other hand, the larger COSMOS clinical endpoint paper did not show a significant reduction in total cancer, cardiovascular disease, or all-cause mortality.^[27]

This is a useful example of the broader theme that biological aging clocks may detect small, integrated shifts across many biological systems. A multivitamin is not a targeted anti-aging drug. It is a broad nutrition intervention that may work by helping correct or buffer small micronutrient inadequacies that accumulate with age.

Dietary patterns, meat, vegetables, and carotenoids

"Vegetables may leave a stronger methylation signal than expected—and objective blood biomarkers like carotenoids may capture that signal better than food questionnaires." Click To Tweet

When people improve their diet, they often lose weight, eat fewer calories, improve insulin sensitivity, and reduce inflammation at the same time. That makes it difficult to know whether clock improvement is due to diet composition itself or to weight loss and downstream metabolic improvements.

Blood carotenoids, which are more objective biomarkers of fruit and vegetable consumption than food-frequency questionnaires, show strong inverse associations with GrimAge and other epigenetic clocks. In postmenopausal women from the Women's Health Initiative, blood carotenoids were inversely associated with GrimAge and related epigenetic aging measures.^[28]

This suggests that plant-rich diets may leave measurable methylation signatures, but the interpretation should be cautious. Carotenoids may reflect fruit and vegetable intake, overall diet quality, lower smoking exposure, lower inflammation, or other correlated health behaviors. Still, the signal is notable because objective nutritional biomarkers can sometimes reveal diet-aging relationships more clearly than self-reported intake.

Does red meat accelerate epigenetic aging?

There is not much convincing evidence that people who eat a lot of red meat age dramatically faster by epigenetic clocks than vegans or omnivores. One short-term diet intervention is often cited because several clock measures improved rapidly in people consuming a vegan diet compared with an omnivorous diet, but the result should not be overinterpreted as a red-meat-specific effect.

In that study, identical twin pairs were randomized to either a healthy vegan diet or a healthy omnivorous diet for eight weeks. The design included two phases: four weeks of delivered meals and four weeks of self-provided meals. The vegan diet excluded animal products. The omnivorous diet included targets for meat, eggs, and dairy. Both groups were coached toward minimally processed foods and balanced meals.^[29]

In the vegan arm, PCGrimAge, PCPhenoAge, and DunedinPACE decreased over eight weeks, while no epigenetic clock or telomere measure significantly changed in the omnivorous cohort.

The twin design is elegant because it controls for genetics, age, and sex within pairs. But interpretation is limited by short duration and potential confounding by reduced calorie intake and weight loss in the vegan group. The result may reflect energy restriction, weight loss, improved diet quality, increased plant-food exposure, or a combination of these factors rather than veganism per se.

Exercise and physical activity

"Exercise is a true"polypill"... but if the intervention is only step count or light daily movement, blood methylation clocks may not move very dramatically." Click To Tweet

Exercise is one of the strongest known lifestyle interventions for longevity, but the epigenetic-clock data are more complicated. Large observational studies using wearables or objective activity measures generally show that more movement is associated with younger epigenetic profiles, but the effect in blood is often modest.

In a large community-based cohort from Bonn, Germany, higher physical activity was associated with lower GrimAge acceleration, especially for daily step counts, total MET-hours, and the percentage of time spent in moderate-to-vigorous physical activity.^[30] The biggest differences appeared between low activity and average-to-moderate activity, with the benefit leveling off at higher activity levels. For example, the difference between 2,300 steps/day and 8,800 steps/day was estimated at roughly 21 months of GrimAge acceleration, and the apparent "maximum" benefit occurred around 11,247 steps/day, 34.7 MET-hours/day, and about 1.5 hours/day of moderate-to-vigorous activity. The signal was strongest for GrimAge and PhenoAge, not the older first-generation Horvath and Hannum clocks. Greater step counts, MET-hours, and moderate-to-vigorous activity were associated with lower PhenoAge acceleration, but the researchers did not find clear associations with Horvath or Hannum age acceleration.

In a large, nationally representative longitudinal study of U.S. adults over age 50, physically active participants had lower epigenetic age acceleration across GrimAge acceleration, PhenoAge acceleration, and DunedinPACE compared with inactive participants.^[31] Physical activity was associated with 1.26 years lower GrimAge acceleration, 1.70 years lower PhenoAge acceleration, and 0.05 lower DunedinPACE, meaning a slower estimated pace of biological aging per chronological year. Both current physical activity and long-term accumulation of physical activity were the strongest predictors of lower epigenetic aging. This suggests two complementary ideas: being active now matters, but maintaining activity over time may also leave a favorable biological imprint.

The somewhat modest effects may be explained by the fact that blood methylation clocks do not capture every tissue-specific adaptation to exercise, especially adaptations in skeletal muscle, heart, vasculature, brain, and mitochondria. Exercise improves cardiovascular fitness, glucose control, body composition, muscle function, vascular health, brain health, inflammation, and disease risk, but to truly impact epigenetic clocks, the exercise stimulus may need to be more deliberate than mere "physical activity."

Deliberate exercise may be a stronger epigenetic lever

"Light movement helps health, but stronger training that improves VO2 max may be more likely to move blood-based aging clocks." Click To Tweet

Blood methylation clocks may under-detect some benefits of exercise unless the intervention is strong enough to produce major physiological adaptation, such as a meaningful improvement in VO2 max.

A six-month cycling-based endurance training intervention tested this idea in physically inactive but generally healthy Belgian adults aged 35–65. Participants completed about 4.5 hours of exercise per week.^[32] VO2 max increased by about 20%, and body composition improved. On average, GrimAge decreased by 7.44 months relative to the expected trajectory of normal aging. GrimAge changes reflected improvements in VO2 max, but not changes in body composition.

This suggests that more intense or structured endurance training may move blood-based aging clocks more clearly than low-intensity movement alone, especially when it produces measurable gains in cardiorespiratory fitness.

Body temperature, metabolism, and hibernation

"Sauna may mimic some exercise-like stress responses. Lower body temperature and hibernation may slow metabolism. Both are interesting, but they are different biological ideas." Click To Tweet

Temperature and metabolism are emerging areas in epigenetic aging research. One mouse study found that stimulating neurons in the preoptic area of the brain lowered body temperature, and the mice whose body temperature was lowered aged more slowly by methylation clocks across multiple organs. This suggests that temperature, metabolism, and biological aging may be mechanistically linked.^[33]

Hibernation biology points in a similar direction. During hibernation, metabolism and physiological activity slow dramatically. In marmots, methylation clocks did not advance during hibernation in the same way they did during active periods. The idea is intuitive: if metabolism, neural firing, immune activity, protein turnover, and other biological processes slow down, then some forms of damage accumulation may also slow.^[34]

A related mouse experiment used chemical and genetic activation of hypothalamic neurons to induce a torpor-like state, a state of reduced activity similar to hibernation. Over four days, body temperature fell by nearly 10°F, or about 7°C; metabolic rate fell by about 56%; and food intake fell by about 81%, with recovery after removal of the stimulus.^[33] After three months, torpor-like mice aged about 80% less in blood and 20% less in liver than controls, with no consistent kidney or cortex effect. After nine months, blood epigenetic age was about three months younger than controls. The authors reported that the effect on blood epigenetic aging was mediated by lower core body temperature rather than caloric restriction or lower metabolic rate.

This is a mechanistic mouse study, not a human intervention. It suggests that core body temperature may influence epigenetic aging rates in certain tissues, especially blood, and helps explain longevity patterns observed in torpor and hibernation biology. None of this means people should try to chronically lower their body temperature or assume cold exposure slows human epigenetic aging.

Can sauna mimic exercise-like signals?

"Heat exposure may be an exercise-like stressor, but we do not yet know whether it produces the same epigenetic aging-clock effects as structured endurance training." Click To Tweet

There is not yet a strong epigenetic-clock literature showing that sauna use reverses biological age. But sauna and hot-water immersion can increase heart rate, induce sweating, raise core temperature, challenge the cardiovascular system, and activate stress-response pathways that overlap with some aspects of exercise physiology.

This fits the framework of heat as a hormetic stressor. Heat exposure produces mild hyperthermia that activates protective stress responses, including heat shock proteins, Nrf2-related antioxidant pathways, inflammatory balancing, neuroprotective pathways, and cardiovascular adaptations.

The current evidence supports sauna as an exercise-like physiological stressor with plausible healthspan relevance, but it should not yet be presented as an epigenetic aging clock intervention.

Sleep disruption

"Sleep should be framed as a high-confidence health behavior but a still-developing epigenetic-clock story." Click To Tweet

Sleep is one of the most intuitive longevity levers, and it is clearly important for health. But the epigenetic-clock literature is still limited compared with diet, supplements, and exercise.

In work using the Women's Health Initiative and other cohorts, people with sleep disruptions showed increased epigenetic age. One study analyzed 2,078 postmenopausal women from the Women's Health Initiative, with an average age of 65 years. Sleep was measured using insomnia symptoms and self-reported sleep duration. Blood DNA methylation was used to estimate intrinsic and extrinsic epigenetic age acceleration, and the authors also estimated immune-cell aging markers, including naive CD8 T cells and late differentiated CD8 T cells.^[35]

Insomnia symptoms were associated with advanced epigenetic age and more late differentiated CD8 T cells, but not with naive T cells. Self-reported short and long sleep duration were not associated with epigenetic age in this study, but short sleep was associated with fewer naive T cells, while neither short nor long sleep related to late differentiated T cells.

This is not surprising, but it is observational, meaning the direction of causality is difficult to establish. Poor sleep could accelerate aging biology, but poor health, pain, depression, stress, obesity, and inflammation could also disrupt sleep and raise clock scores.^[35]

There are several plausible mechanisms by which chronic sleep disruption could accelerate aging: increased inflammation, altered appetite, weight gain, metabolic dysfunction, circadian disruption, and broader physiological stress. With modern wearables, researchers can now examine deep sleep, REM sleep, sleep regularity, sleep timing, and sleep fragmentation in relation to GrimAge, PhenoAge, and DunedinPACE.

The practical takeaway is that sleep should be treated as a high-confidence health behavior, but the epigenetic-clock evidence remains early and mostly observational.

Social connection and loneliness as age accelerators

"Social connection may be a longevity-relevant factor, tracking with lower inflammation and slower epigenetic aging." Click To Tweet

Loneliness is widely recognized as a major risk factor in older adults, sometimes compared in magnitude to smoking. Social connection also appears to show up in methylation data.

A study examining cumulative social advantage—a measure of social connectedness, community ties, relationships, and social resources—in relation to biological aging found that this factor was associated with slower epigenetic aging and lower systemic inflammation.^[36]

The analysis used data from 2,117 adults in the MIDUS biomarker cohorts. Cumulative social advantage captured social connection across family relationships, religious and faith-based support, emotional support, and community engagement. Higher cumulative social advantage was associated with slower epigenetic aging, especially GrimAge and DunedinPACE, and with lower IL-6.

The authors did not find significant associations with urinary cortisol, cortisone, or catecholamines. This pattern fits the idea that stable social resources may be biologically embedded more through chronic inflammatory burden and cumulative aging signatures than through single overnight neuroendocrine readings.

Social connection should be considered part of the broader longevity environment. It may not be a "biohack," but it appears to be biologically legible.

Medical interventions

Some of the strongest epigenetic aging reversal signals come from medical contexts where a disease-related aging accelerator is treated. These examples are important because they clarify what large clock reversals often represent: not magic rejuvenation, but the removal or control of a biological stressor.

HIV and antiretroviral therapy

HIV is one of the clearest examples of a condition associated with epigenetic age acceleration. HIV-positive people can show several years of accelerated epigenetic age in blood, and adherence to antiretroviral therapy can reverse a meaningful portion of that acceleration.^[37]

This does not mean antiretroviral therapy is an anti-aging intervention for people without HIV. It means that treating a chronic viral and inflammatory stressor can reduce a disease-associated aging signal.

Anti-TNF therapy

Anti-TNF therapy is another medical intervention with evidence for favorable clock effects, especially in inflammatory disease contexts such as moderate-to-severe psoriasis.^[38]

This fits the broader principle that chronic inflammation can accelerate aging-clock measures, and suppressing that inflammation in people with inflammatory disease may partially normalize the signal.

Metformin

Metformin belongs in the "modest or uncertain clock signal" bucket, not the robust-reversal bucket. A randomized epigenetic aging analysis in overweight or obese postmenopausal breast cancer survivors found no significant epigenetic aging differences across metformin, weight loss, combined treatment, or placebo over six months.^[39]

This does not rule out potential benefits of metformin in specific populations, but it argues against treating metformin as a reliably proven epigenetic age-reversal intervention.

GLP-1 therapy and semaglutide

GLP-1 therapy is promising but preliminary for epigenetic aging, especially because weight loss, inflammation reduction, and improved metabolic health may all contribute to clock changes.

In an analysis of a 32-week randomized, double-blind, placebo-controlled trial in people with HIV-associated lipohypertrophy, participants were randomized to weekly semaglutide or placebo. Semaglutide was associated with lower values across multiple clocks:

• PCGrimAge by -3.1 years
• GrimAge v1 by -1.4 years
• GrimAge v2 by -2.3 years
• PhenoAge by -4.9 years
• DunedinPACE by -0.09 units, interpreted as approximately 9% slower pace. Eleven organ-system clocks reportedly decreased, most prominently inflammation, brain, and heart.^[40]

This is an important early signal, but it should be interpreted carefully. The participants had a specific metabolic and inflammatory condition, and the intervention likely affected weight, insulin sensitivity, liver fat, inflammation, and adipose distribution simultaneously. The result may tell us less about GLP-1 drugs as direct "anti-aging" agents and more about the extent to which correcting metabolic dysfunction can move epigenetic aging clocks.

Can AI build better aging clocks?

"AI may build stronger clocks, but the field still needs validation against interventions, disease outcomes, mortality, and functional health." Click To Tweet

Artificial intelligence may help build better aging clocks, but the central challenge remains the same: a clock is only as useful as the outcomes it predicts and the intervention responses it has been validated against.

GrimAge was published in 2019, before the current AI boom, and yet it remains one of the strongest methylation-based predictors of mortality. That is a useful reminder that complexity alone does not make a clock better. The best clocks are not simply the most advanced computational models; they are the models that predict disease, functional decline, mortality, and meaningful changes in response to interventions.

Newer AI-based and multi-layered clocks are beginning to appear, including systems such as AltumAge, SystemsAge, OMICmAge, and next-generation GrimAge-family clocks. The hope is that one or more of these newer tools will outperform current clocks, especially for clinical trials, where the field needs sensitive biomarkers that can detect whether an intervention changes aging biology before decades of disease or mortality follow-up have passed.

The field is also moving away from the idea that one global biological-age number can capture everything. Future tools are more likely to function as biological dashboards that integrate methylation, proteomics, metabolomics, organ function, imaging, blood biomarkers, electronic health records, and wearable data. AI may be especially useful for integrating these layers, but validation remains the bottleneck.

AltumAge

AltumAge illustrates what deep learning can add to methylation clocks. It is a pan-tissue DNA methylation clock trained with a neural network across a large number of methylation datasets, allowing it to capture more complex and nonlinear relationships between methylation sites. In one analysis, AltumAge performed better in difficult tissues and in older ages, where earlier clocks often underestimate age. It also showed biologically plausible age acceleration in tumors, high-passage cells, immune dysfunction, mitochondrial dysfunction, multiple sclerosis, type 2 diabetes, HIV, and other disease contexts.^[41]

AltumAge is therefore a strong example of how AI may improve chronological-age prediction, cross-tissue generalization, and nonlinear modeling. But it should not be confused with a mortality-risk clock like GrimAge. A clock can be excellent at estimating chronological age without being the best predictor of mortality, disease, or intervention response.

SystemsAge

SystemsAge represents a different conceptual advance. Instead of asking whether the whole person is biologically "older" or "younger," it estimates aging across 11 physiological systems from a single blood methylation test: heart, lung, kidney, liver, brain, immune, inflammatory, blood, musculoskeletal, hormone, and metabolic systems. In validation work, system-specific scores outperformed existing global clocks for predicting relevant diseases and aging phenotypes. A composite Systems Age score also captured multisystem aging and helped identify distinct biological aging subtypes, each linked to different patterns of health decline and disease risk.^[42]

This is important because aging is heterogeneous. A person may not be globally old or young. They may have relatively faster immune aging, metabolic aging, kidney aging, inflammatory aging, or musculoskeletal aging. A system-specific clock could eventually make biological-age testing more clinically useful by identifying which systems appear most vulnerable and which interventions are most relevant.

OMICmAge

OMICmAge takes another approach: it is a multi-omics-informed methylation clock. It uses DNA methylation to encode information from proteomic, metabolomic, and clinical domains, while still remaining measurable from DNA methylation alone. The goal is to make methylation clocks less like isolated epigenetic readouts and more like integrated biological-risk models.

In validation analyses, OMICmAge showed strong associations with chronic disease and mortality. In one analysis, it had the highest hazard ratios for incident type 2 diabetes, cardiovascular disease, and all-cause mortality among the studied aging biomarkers. In Generation Scotland, OMICmAge ranked second for 5-year and 10-year mortality prediction, just behind PCGrimAge.^[43]

OMICmAge may be one of the most relevant next-generation clocks because it tries to solve a central limitation of methylation-only models: methylation is powerful, but aging is multi-layered. OMICmAge uses methylation as an accessible substrate while incorporating signals that reflect proteins, metabolites, clinical labs, and electronic medical-record-derived risk.

GrimAge v2/v3

Next-generation GrimAge-family clocks build on a more familiar principle: use DNA methylation to estimate mortality-relevant physiology, then combine those signals into a mortality-risk-oriented biological age measure. GrimAge version 2 added methylation surrogates for high-sensitivity C-reactive protein and HbA1c, expanding the model's ability to capture inflammatory and metabolic risk. GrimAge2 outperformed the original GrimAge for mortality prediction across multiple racial and ethnic groups and also predicted or associated with several healthspan-relevant outcomes, including coronary heart disease, lung function, fatty liver, visceral adiposity, waist-to-hip ratio, insulin resistance, and metabolic syndrome-related features.^[28]

The promise of AI-built clocks is not simply that they will generate more impressive biological-age numbers. The promise is that they may help aging biomarkers become more specific, more predictive, and more clinically actionable. A future biological aging report may not say only, "Your biological age is 52." It may show that inflammatory aging, kidney aging, metabolic aging, or brain-associated aging is moving faster than expected, while other systems appear preserved.

That would be a major advance. But the same caution still applies: new clocks need broad validation across populations, tissues, diseases, interventions, functional outcomes, and mortality.

Consumer biological age tests

"Use epigenetic-age testing as a motivation and tracking tool—not as a diagnosis, a death-date estimate, or proof that one supplement"reversed aging." Click To Tweet

Biological-age testing has moved from research labs into consumer wellness. That creates both opportunity and confusion. These tests should not be treated as diagnosis tools, death-date predictors, or proof that a single intervention "reversed aging." Consumer tests may be useful because they can improve motivation and adherence.

The downside is cost. Current tests often cost several hundred dollars. Much cheaper tests are technologically possible, perhaps around $50, but cheaper tests may use different platforms and newer clocks that have not been characterized as extensively in the research literature. Cheaper technology may increase access, but the best-studied clocks are built on platforms like Illumina methylation arrays, which have a long research history and standardized preprocessing pipelines.

Consumer Test List

Epigenetic-age tests can be motivating, but they are optional. If you buy one, choose based on the report you want to follow over time: a broad methylation panel, a named clock such as DunedinPACE or GrimAge, or a simpler consumer score.

What to Look For Before Buying

• Named clocks that use a validated methylation platform: if you want to follow the clocks discussed in this episode, look for reports that explicitly name DunedinPACE, GrimAge / PCGrimAge, PhenoAge / PCPhenoAge, or Horvath / PCHorvath2013. Ideally, a testing service will report multiple clocks.
• Raw data access: if possible, choose a provider that lets you download raw methylation data or IDAT files, so the same sample data can be reanalyzed with newer tools later.
• Specifies the sample type (i.e., blood, saliva, cheek, or another tissue).
• Reports test-retest reliability or technical variation.

The best approach is not to compare one saliva test from one company with one blood test from another and treat the difference as biological change. Instead:

• establish a baseline when healthy and stable;
• use the same company, platform, and sample type for repeat testing;
• retest after enough time for a real intervention to plausibly matter;
• interpret the result alongside conventional biomarkers and functional measures.

Best Broad Option

• TruDiagnostic TruAge - the strongest consumer-facing option if you want a broad methylation report instead of a single biological-age number. The TruAge ecosystem includes DunedinPACE, OMICmAge, organ-system ages, DNAm telomere length, immune/inflammation scores, and smoking/alcohol impact; TruDiagnostic's bioinformatics page also lists legacy PC-clock outputs such as PCHorvath2013, PCPhenoAge, and PCGrimAge.

More Specific Options

• NOVOS Age - an at-home blood test centered on DunedinPACE, organ-age algorithms, biological age, and telomere length. Best fit for consumers most interested in the pace-of-aging and organ-age.
• myDNAge / Zymo DNAge - a targeted sequencing test based on Steve Horvath's original epigenetic aging clock. Best fit for the original Horvath-clock lineage.
• Clock Foundation GrimAge Analysis - a reanalysis option for people who already have raw methylation data / IDAT files and specifically want GrimAge output.

Simple Consumer Scores

• Elysium Index - a saliva-based methylation test that reports biological age, cumulative rate of aging, and nine system ages.
• TallyAge - a cheek-swab test with a simple epigenetic-age score and lifestyle-oriented tracking.

Disclaimer: Why two epigenetic age tests might disagree

Test variability is one of the biggest issues for consumers. Principal component-based clocks such as PC-GrimAge are highly reproducible. If someone measured PC-GrimAge on Monday and again on Wednesday, with nothing meaningful changing biologically, they would expect technical variation on the order of a few months. DunedinPACE is slightly less robust but still highly reproducible.

But reproducibility within a clock is not the same as agreement across clocks. GrimAge, PhenoAge, DunedinPACE, and the original Horvath pan-tissue clock measure different aspects of biology. A person can therefore receive discrepant results without one test being "wrong." The Horvath pan-tissue clock may be more useful for stem-cell biology or cross-tissue questions, while GrimAge is stronger for mortality risk and DunedinPACE is designed to estimate the current pace of aging.

Conclusion

Epigenetic aging clocks have changed the way scientists study aging by making it possible to measure biological signals over months or years rather than waiting decades for disease or mortality outcomes. They can reveal whether a person's methylation profile looks older, younger, faster-moving, or more disease-prone than expected—but they are not all measuring the same thing, and no single clock captures the entire aging process.

The most useful way to interpret these clocks is as biological integrators. They can reflect smoking, inflammation, metabolic health, immune aging, nutritional status, organ stress, and long-term exposure history. But they can also miss important forms of damage, and a favorable clock change does not automatically mean a person has gained years of life. The field still needs deeper validation against clinical outcomes, functional health, and intervention response.

For now, epigenetic clocks are best understood as powerful research tools and emerging clinical-adjacent biomarkers—not definitive report cards on aging. Their promise is not that they reduce aging to one number, but that they may help reveal which biological systems are changing fastest, which interventions are working, and how aging can be measured more precisely in humans.