# Biological aging and epigenetic clocks: what a 2026 Nature Medicine study says about diet and exercise > Researchers pooled 51 human trials and 16 DNA clocks. Diet with exercise often lowered scores; supplements did less. A lower clock is not proof you live longer. Published 2026-11-14 · Updated 2026-11-14 · revns editorial If you train and eat well, a major **2026** analysis says your habits may already be moving a research marker of aging in the right direction. In *Nature Medicine*, **Raghav Sehgal**, **Albert Higgins-Chen** and colleagues at Yale report on **TranslAGE**, a harmonized look at **51 longitudinal human trials** that measured DNA methylation before and after an intervention. **Diet combined with exercise** and several **lifestyle** programs were among the approaches linked to **lower epigenetic age** on multiple [epigenetic clocks](/dictionary/epigenetic-clock). **Over-the-counter supplements** in the database, on average, showed **weaker or inconsistent** shifts. The authors repeat a crucial limit throughout: **a change on a clock is not proof you will live longer**, and consumers should not treat mail-order tests as medical guidance. **TL;DR** | Topic | What TranslAGE reports | What it means for you now | | --- | --- | --- | | Design | 51 human intervention studies with blood DNA methylation reprocessed the same way | Research roadmap, not new clinical rules | | Clocks | 16 prominent epigenetic clocks plus 94 related biomarkers per study | Different clocks can disagree on the same trial | | Lifestyle | 15 lifestyle trials; DunedinPACE decreased significantly in 8 | Diet plus movement remain the clearest DIY levers | | Supplements | Supplements category showed fewer significant clock shifts than pharmacological or lifestyle groups in pooled tests | Skip buying hope in a bottle; food and training come first | | Drugs in dataset | Trials included metformin, semaglutide, anti-TNF therapy and others | Reported for science, not self-prescribing | | Bottom line | 19 of 51 interventions lowered clocks in at least one model (13 after multiple-testing correction) | Keep WHO-aligned activity and food patterns; ignore aging-test hype | ## What is an epigenetic clock? An **[epigenetic clock](/dictionary/epigenetic-clock)** estimates **biological age** from chemical tags on DNA, mostly **methylation** patterns in blood. The tags change with age and, in some trials, shift after diet, exercise, illness or medication. Algorithms such as **Horvath**, **PhenoAge**, **GrimAge** and **DunedinPACE** translate those patterns into a single score or a **pace of aging** number. Clocks correlate with health risks in population studies, which is why longevity researchers want faster feedback than decades-long survival trials. They are **not** the same as a birthday. Two people born the same year can read different clock ages. A lower score after a month of salads does not automatically mean your arteries are younger. ScienceDaily and Yale University covered the paper on **13 November 2026**, pointing readers to the August 2026 journal publication. The press framing highlights **measurable** responses where past reviews argued clocks were too noisy to trust. Higgins-Chen’s team still insists the **clinical meaning** of a drop remains unfinished science. ## What did the Nature Medicine study actually do? Instead of running one new trial, the group **re-analyzed public and private methylation datasets** from longevity-themed studies. They built **TranslAGE** with consistent metadata: age, sex, follow-up time, disease status and intervention type. Every sample passed through the same pipeline (**MethylCIPHER 2.0**) so effect sizes could be compared across trials. They focused on **whole blood** DNA. That matters for trainers: clocks here reflect immune and blood-cell biology, not muscle fibre changes directly. ### Control checks To see whether the pipeline could spot real signals, the authors included **four control datasets** and **nine pro-aging event datasets** (events expected to raise biological age markers). In a one-sample t-test across 16 clocks, **pro-aging events** averaged about **+0.14** standardized effect size and **longevity-style interventions** about **−0.06**, both **P < 0.0001** versus a zero line. Paired comparisons between categories were also significant at **P < 0.001**. Those numbers describe statistical separation inside the database, not how many years you gain. ### How many interventions moved the clocks? Across all 51 interventions and 16 clocks, **19 interventions significantly decreased** epigenetic age on at least one biomarker (**13** after multiple-testing correction). **Five increased** age markers (**3** after correction). **26** showed no significant pooled effect. So the majority of tested ideas in the catalog failed to move clocks under this strict panel. Interventions were grouped into **lifestyle** (diet and exercise programs), **pharmacological**, **supplements** and **medical procedures**. A category-level test found **more significant biomarker shifts in pharmacological trials** than in supplement or procedure groups, which matches the headline that pills in the dataset often moved needles hard. That does **not** mean you should seek prescriptions to chase a clock score. Those trials enrolled specific patients under medical supervision. For everyday readers, the lifestyle column is the actionable one. ## Did diet and exercise slow biological aging in the data? **Yes, often, in this harmonized set**, especially when trials combined eating changes with activity. Lifestyle trials in TranslAGE included **Mediterranean**, **low-fat**, **low-carbohydrate** and **plant-forward** eating plans, sometimes with **calorie targets**, paired with **aerobic training**, **resistance training** or **mixed programs**. ScienceDaily quoted lead author Sehgal saying that **regular physical activity plus a healthy eating plan consistently reduced epigenetic age** across the clocks they tested. In the paper’s lifestyle-only breakdown, **DunedinPACE** (a pace-of-aging clock) decreased significantly in **8 of 15** lifestyle interventions—the highest count in that subset. **SystemsAge** decreased in **6**, and **DunedinPoAm38** in **5**. One-way tests on effect sizes ranked **DunedinPoAm38**, **DunedinPACE** and **GrimAgeV2** among the largest downward shifts for lifestyle, with several passing multiple-testing correction. Those statistics describe **consistency across trials**, not a single magic workout. Some lifestyle studies were small. Durations ranged from weeks to years. Populations mixed healthy volunteers and people with metabolic disease; **disease status** strongly predicted how much clocks moved, with larger shifts often seen when participants already had a condition. None of that replaces public-health basics you can act on without a lab kit. ### What fits WHO and CDC guidance? The World Health Organization still recommends adults accumulate **150 to 300 minutes** of **moderate** aerobic activity per week, or **75 to 150 minutes** of vigorous work, plus **muscle-strengthening** on **two or more days**. Walking counts. So do squats, rows and presses. Diet patterns in TranslAGE overlap with Mediterranean-style eating: **vegetables, legumes, nuts, whole grains**, olive oil or other unsaturated fats, and **protein** spread across the day. Our guide on [how much protein you need](/blog/how-much-protein-per-day) covers vegetarians, vegans and non-vegetarian plates without centreing meat. For steady aerobic work, [zone 2 training](/blog/zone-2-training-explained) explains how easy conversational pace links to the moderate band. For step targets, see [how many steps per day you actually need](/blog/how-many-steps-per-day-do-you-need). ## What about supplements and “anti-aging” products? Headlines from Yale stress disappointment for **popular supplements**. In TranslAGE’s supplement category, examples include **omega-3 fatty acids** and **folate** trials. Pooled analyses showed **fewer clocks with significant decreases** than in pharmacological or many lifestyle studies. Individual supplement trials sometimes moved one biomarker, but results were **less reproducible across clocks and study repeats**. The authors propose **two rules** for believing an intervention truly affects aging biomarkers: the same intervention should move **multiple related clocks in one study**, and a **second study of the same intervention** should move the **same clocks** in the same direction. Many supplement entries failed those replication checks. That aligns with broader nutrition science: **whole-food patterns** outperform isolated molecules in long-term health trials, and single supplement studies often disagree. It does **not** prove every bottle is useless, but it is a reason to **spend money on groceries and gym time first**. Avoid influencer stacks that promise to “reverse” methylation age in 30 days. If a product cites this paper, read whether the citation is about **research clocks**, not FDA-approved outcomes. ## Why does the paper mention prescription drugs? Pharmacological trials in the database include **metformin**, **semaglutide**, **rapamycin**, **ketamine** and **anti-TNF biologics**, among others. In category comparisons, pharmaceutical interventions produced **more significant biomarker hits** than supplements on average, and some drugs showed **large downward effect sizes** on mortality-trained clocks. This run’s rules exclude **promoting weight-loss or GLP-1 drugs** as lifestyle hacks. Report them because they appeared in **published trials the authors re-analyzed**, not because readers should request them for aging. Metformin and semaglutide treat **diabetes and obesity** under medical supervision. Anti-TNF drugs manage **inflammatory diseases**. None are anti-aging prescriptions for healthy gym-goers. If you have diabetes, inflammatory arthritis or obesity-related risks, those decisions belong with a clinician who knows your labs and medications. Clock shifts in a trial do not add a new indication. ## Can you trust a consumer biological-age test? Companies already sell **epigenetic age** and **pace-of-aging** reports from mailed blood or saliva kits. TranslAGE is aimed at **trial designers** choosing **which clock to monitor** and **how long to run a study**, not at shoppers. The same Yale group and others have warned that **different clocks disagree** on whether an intervention “worked.” In [Yale News](https://news.yale.edu/2026/08/21/can-you-slow-aging-process-study-reveals-which-interventions-might-help), Higgins-Chen said researchers still need to learn **what a clock change means for long-term disease risk and functional decline** before treating clocks like validated trial endpoints. The Nature Medicine authors frame that as the gap between **prognostic** clocks and **surrogate** endpoints in trials. For training, cheaper proxies still help: - **Resting heart rate** trending down with fitness - **Step count** and **moderate minutes** versus your usual week - **Strength numbers** progressing under [progressive overload](/blog/progressive-overload-in-plain-words) - **Sleep regularity** and duration The revns guide [how fit am I for my age](/blog/how-fit-am-i-for-my-age-sleep-steps-training-hrv) compares sleep, steps, training frequency, resting heart rate, VO2 max estimates and HRV to typical ranges by decade. Those metrics track behaviours, not methylation. revns logs workouts and sleep-related fields you enter. **Apple Health, Strava and Garmin sync are on the roadmap, not live yet.** Nothing in the app interprets epigenetic clocks. ## What to do this week 1. **Train on schedule**: at least **two** strength sessions and **150+ minutes** of moderate cardio (or the vigorous equivalent), matching your recovery. After 40, the [strength after 40 guide](/blog/strength-training-after-40-what-changes-and-what-to-do) covers joint-friendly progression. 2. **Plate method for food**: half vegetables or fruit, a quarter protein (tofu, dal, eggs, fish, poultry or lean meat as you eat), a quarter whole grains or starchy plants. Add water across the day; see [daily water intake](/blog/how-much-water-should-you-drink-daily). 3. **Sleep anchor**: same wake time seven days a week when you can; cut late caffeine. Poor sleep raises inflammation markers independent of clocks. 4. **Log a baseline week**: steps, sessions, resting heart rate and mood in one place, then compare in four weeks. The [five-number weekly check-in](/blog/five-numbers-weekly-health-check-in) lists fields worth tracking. 5. **Ignore clock FOMO**: if a mail-order test upsets you more than it helps, stop testing. Behaviour beats biomarker anxiety. ## When should you see a professional? Talk to a doctor or qualified clinician before a big training jump if you have heart disease, uncontrolled blood pressure, diabetes, kidney disease, cancer treatment, or joint issues that limit load. Seek urgent care for **chest pain**, **fainting**, **one-sided weakness**, **sudden confusion**, or **severe headache**. Book a routine visit for **unplanned weight loss**, **persistent fatigue** despite sleep, **night sweats**, or **bruising** you cannot explain. Mental health counts too. If low mood, hopelessness or anxiety lasts more than two weeks, or you might harm yourself, contact a clinician or crisis line now. In the US, call or text **988**. In the UK and Ireland, Samaritans are on **116 123**. In India, KIRAN is on **1800-599-0019**. Other countries: [findahelpline.com](https://findahelpline.com). ## Related - [Strength training after 40](/blog/strength-training-after-40-what-changes-and-what-to-do) - [How fit am I for my age?](/blog/how-fit-am-i-for-my-age-sleep-steps-training-hrv) - [Five numbers worth a weekly health check-in](/blog/five-numbers-weekly-health-check-in) - [VO2 max: what it is and how to raise it](/blog/vo2-max-what-it-is-and-how-to-raise-it) - [Heart rate zone calculator](/tools/heart-rate-zone-calculator) - [Multi-sport training log](/features/multi-sport-tracker) This is general training guidance, not medical advice. ## Sources - [Sehgal R, Higgins-Chen A et al. Responsiveness of epigenetic aging biomarkers to longevity interventions in humans. Nature Medicine 2026](https://doi.org/10.1038/s41591-026-04562-9) - [Yale University. Yale scientists reveal what really slows aging and what doesn't. ScienceDaily, 13 November 2026](https://www.sciencedaily.com/releases/2026/11/261112230452.htm) - [Yale School of Medicine. Yale researchers create roadmap to measure whether longevity interventions work](https://medicine.yale.edu/news-article/yale-researchers-create-roadmap-to-measure-whether-longevity-interventions-work/) - [Bull FC et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med 2020](https://doi.org/10.1136/bjsports-2020-102955) ## Questions ### Can a blood test tell me my biological age today? Consumer and research labs sell epigenetic age estimates from blood DNA methylation patterns. They can be interesting for tracking, but they are not regulated like clinical tests and they do not diagnose disease. The 2026 Nature Medicine paper treats clocks as research biomarkers, not endpoints your doctor will order to decide care. ### Did the study prove that diet and exercise make people live longer? No. The Yale-led team harmonized 51 existing human trials and asked which interventions shifted 16 epigenetic clocks and 94 related DNA methylation measures. Lifestyle trials often moved clocks in the expected direction, but the authors stress that a smaller clock score does not yet mean lower risk of heart disease, dementia or death. ### Should I take anti-aging supplements because this study says they fail? Do not start or stop supplements based on this article. The pooled data reported weaker clock shifts for the supplement category than for combined diet-and-exercise trials, but individual products varied and the database mixed many small studies. A balanced diet and regular activity remain the evidence-backed habits for active adults. ### Why mention prescription drugs at all? The paper includes pharmacological trials such as metformin and semaglutide because those drugs appeared in the underlying clinical studies. The analysis is for researchers comparing biomarkers, not a guide for prescribing. Only a clinician who knows your history can decide whether a drug is appropriate. ### Which clocks were most useful in the analysis? Clocks trained to predict mortality or pace of aging, including newer models such as DunedinPACE and GrimAgeV2, responded more consistently than first-generation chronological-age clocks in the harmonized dataset. Even so, different clocks sometimes disagreed on the same trial, which is why the authors warn against trusting a single score. ### What should an active adult do this week? Keep training, eating enough protein and fibre, and sleeping on a steady schedule. If you want numbers alongside effort, log steps, sessions and resting heart rate rather than chasing a mail-order aging score. See a clinician for chest pain, fainting, sudden weakness or unexplained weight loss. --- Canonical: https://revns.com/blog/epigenetic-clocks-diet-exercise-nature-medicine-2026 Index of everything: https://revns.com/llms.txt revns gives training guidance, not medical advice.