What 51 Longevity Studies Reveal About Drugs, Diet and Biological Aging

Those findings should not be read as evidence that semaglutide is an approved anti-aging treatment. The semaglutide evidence comes from specific study populations and follow-up periods, and the measured changes do not by themselves establish improved survival or healthspan.

The supplied reporting also describes anti-TNF therapies as producing favorable signals. That result should be treated cautiously here because the available source record does not provide the underlying study’s full methods or effect estimates for that intervention category.

Diet plus exercise

Healthy eating combined with exercise also showed a favorable pattern. The signal reportedly appeared across Mediterranean, low-fat and low-carbohydrate dietary formats.

That pattern suggests the broader combination of nutritional quality and physical activity may matter more than treating one named diet as uniquely “anti-aging.” It does not mean that all versions of these diets produce the same outcome, nor does it establish that a particular fasting schedule or workout plan will reduce anyone’s biological age by a fixed number of years.

Over-the-counter supplements

Most over-the-counter supplements showed little measurable effect in the synthesis.

This is a useful corrective to the idea that longevity supplements have already been shown to reset biological age. A supplement can have other effects—or could require a different dose, duration, tissue or biomarker to detect an effect—but a change in marketing claims is not the same as a validated change in health outcomes.

Medical procedures

Some medical procedures also produced changes in methylation-based biomarkers. However, the meaning of those changes depends on the procedure, the people studied, the tissue sampled, the clock used and the length of follow-up. A biomarker response alone cannot establish that a procedure durably slows aging or improves long-term function.

How this differs from Dean Ho’s DELTA experiment

The DELTA study took a very different approach. It was an N=1 digital-health protocol in which Dean Ho served as the sole participant. The intervention package included extended daily fasting, exercise, additional sleep and a Mediterranean-style diet. Using the study’s April 2025 measurements, the team reported an estimated biological age of about 31.8 years for a 47-year-old participant.

That result is striking, but it cannot answer the same question as a multi-study synthesis. Because several behaviors changed together, DELTA cannot determine whether fasting, exercise, sleep, diet, weight change, another factor or the combination produced the result. A single participant also cannot establish an average effect for other people.

The AI-estimated age should therefore be understood as an output of a biomarker model, not as proof that Ho had physically become 15 years younger or that his lifespan had increased. The DELTA paper presents the work as an individualized digital-health and dynamic-biomarker protocol, which is valuable for generating hypotheses but limited for generalization.

Why biomarker responsiveness matters—but is not enough

The Yale-linked analysis addresses an important problem in aging research: many proposed interventions are difficult to test using outcomes such as disability, dementia, cardiovascular disease or mortality because those outcomes may take years or decades to appear.

If DNA-methylation biomarkers can be shown to respond reliably and predict meaningful clinical outcomes, they could help researchers screen promising therapies over months or a few years. That could make longevity and chronic-disease trials faster and less expensive.

But responsiveness is only the first requirement. Before these markers can serve as dependable surrogate endpoints, researchers need to establish that changes are:

• technically reliable when the same sample is measured repeatedly;
• reproducible across laboratories, populations and tissues;
• consistent across relevant epigenetic clocks rather than limited to one model; and
• predictive of outcomes people care about, such as disease risk, physical function, healthspan or survival.

Those safeguards are especially important because technical noise can produce large differences between repeated measurements. One study found deviations of up to nine years between replicate measurements for several prominent epigenetic clocks, while principal-component versions improved agreement to within about 1.5 years.

The bottom line

The 51-study analysis supports a careful conclusion: some established medications and combined lifestyle interventions appear capable of shifting DNA-methylation measures associated with biological aging, while most over-the-counter supplements do not show much effect in the available synthesis.

It does not show that metformin, semaglutide, anti-TNF therapy, a particular diet or any procedure reverses aging, guarantees better health or extends life. DELTA’s estimated 32-year biological age is an intriguing individualized observation, but its N=1 design makes it much weaker evidence for general claims.

For now, the most important advance is methodological: researchers may be getting better tools for detecting whether interventions change aging-related biology. The next test is whether those molecular changes reliably translate into longer, healthier lives.