Dr. Steve Horvath is a biostatistician and researcher best known for discovering “epigenetic clocks,” mathematical models that estimate a person’s biological age by reading chemical tags known as DNA methylation. While your birthday counts how many years you’ve been alive, Horvath’s original clock, as well as the many biological age clocks inspired by it, reflects how fast your body is really aging. These clocks have opened a new window into the biology of getting older, revealing that people with different diseases, lifestyles, or treatments can look biologically older or younger than their calendar age. Here, Dr. Horvath reflects on how this idea emerged, what epigenetic clocks can (and can’t) tell us about health, and how they might reshape the future of aging research.
1. Before you built your first epigenetic clock, did you actually expect there to be a “clock” hiding in DNA methylation, or did the data surprise you?
It was a complete surprise. I had been working hard on building age predictors based on gene expression data. When I turned to DNA methylation, I did not expect to find anything as striking as what emerged. The accuracy was extraordinary: the predictions matched people’s ages with a Pearson correlation above 0.9, which in this context means an extremely tight relationship between predicted and actual age across a broad spectrum of ages.
I spent months calling these things “age predictors” or “biomarkers” before settling on the term “epigenetic clock.” I felt the exceptional accuracy earned the label: these weren’t rough estimates, they were clocks. I still distinguish the term “biomarker of aging” from that of clocks (defined as very accurate age predictors).
2. You were originally trained as a mathematician and biostatistician. How has that background shaped the way you think about biology and aging?
It was both a blessing and a curse. A blessing because it allowed me to see the signal through the noise. When I first looked at DNA methylation data in the context of aging, I immediately recognized that this was my future. I dropped my gene expression project without hesitation. And methylation never disappointed: it led to a series of increasingly powerful clocks, culminating in the pan-mammalian clocks that can measure age across all mammalian species. But my training was also a limitation. It took me many years to learn enough molecular biology and clinical medicine to put the clocks into proper biological and clinical context. Studying math and biostatistics is not the most direct path to becoming a geroscientist. Medicine or molecular biology may be more efficient routes for many students. Or focus on computational biology. Honestly, I would tell students: don’t follow my exact tracks. And don’t spend too much time in graduate school.
3. When researchers apply epigenetic clocks to almost any disease group (heart disease, diabetes, HIV, neurodegenerative diseases) they often see signs of accelerated epigenetic aging. Does this make you think that diseases are different faces of the same aging process? Or do you see disease and aging as separate processes that just happen to overlap in the clocks?
The situation is more complicated than either framing suggests. You have to be careful about which epigenetic clock is being used and which tissue is being analyzed. Clocks trained on bulk tissue capture both changes in cell composition (which cell types are present) and cell-intrinsic changes (what is happening inside each cell). These are very different biological signals. And clocks applied to blood may give a completely different picture from clocks applied to specific organs. For example, epigenetic age in the brain may bear little relation to epigenetic age in blood.
Overall, I view disease as fundamentally different from epigenetic aging. When an overlap is observed, the important question is: where exactly is the overlap? In which cell type? According to which clock? Without that precision, the apparent connection between disease and aging can be misleading. Only a subset of DNA methylation changes are actually causally involved in organ dysfunction.
4. You’ve helped develop clocks not just for humans but across many different species. What have those pan-mammalian clocks taught you about what is universal in aging, and what might be uniquely human?
I think the most stunning insight is that one can measure age across all mammalian species using several hundred CpG sites located in stretches of DNA that are almost identical in every mammal (what scientists call “ultraconserved” regions). That finding shows that specific aspects of epigenetic aging are deeply conserved.
But I want to avoid oversimplification. Millions of positions on the DNA molecule show age-related changes that are specific to a particular tissue type and species. Methylation changes in enhancer regions, for instance, are often tissue-specific and species-specific. So the picture is nuanced: a core conserved aging program exists alongside enormous species-specific and tissue-specific variation.
5. There are now many different epigenetic clocks out there, and sometimes they don’t agree with the same person or intervention. When clocks disagree, how do you interpret that? What do you pay attention to first?
Millions of positions on the DNA molecule change with age, and you can build clocks with very different properties depending on which positions you select and what you train them to predict. Some clocks are designed to be conserved across tissues and species, like the pan-mammalian clocks. Others are tailor-made for a specific cell type, such as human neurons. Clocks can be developed to capture specific facets of aging biology.
When two clocks disagree, it usually means they are measuring different things, which is actually informative. The current frontier is building blood-based methylation clocks that reflect the age of specific organs: a blood test that tells you the age of your liver, your kidneys, or your lungs. Rigorous clinical testing will determine which of these clocks ends up being genuinely useful in the clinic. A lot of work remains but I can see the promised land very clearly.
6. If you could correct one common misconception the public has about aging, what would it be?
That aging is something we just have to accept. For most of human history, that was true. We had no way to even measure aging precisely, let alone intervene. But now we can measure it, and measuring is always the first step toward changing something. The aging biomarker community builds better and better aging clocks. I have watched the field go from skepticism about epigenetic clocks to serious drug development programs in barely a decade. We are not helpless in the face of aging anymore. That, to me, is the most exciting thing happening in science right now. It is more hopeful than accepting the status quo and the inevitability of aging.
Lifespan Magazine is proud to support medical research and the Lifespan Foundation
Last Updated: Oct 19, 2026 · 5 minute read
