A little over a decade ago, scientists found something that sounded almost too good to be true. By reading the biochemical tags on a person’s DNA, they could estimate that person’s age, often within a few years, without knowing their birthday. These “epigenetic aging clocks” became one of the most talked-about ideas in modern biology, and deservedly so. For the first time we had a biological measure of aging instead of a number on a driver’s license.
But aging wasn’t the only health insight that epigenetics could capture. If the epigenome encodes something as subtle as the pace of your aging, it almost certainly encodes a lot more. The body’s most dynamic data layer could be interpreted to learn new dimensions like nutrition, inflammation, and disease risk too.
A quick refresher on what an aging clock measures
Besides accumulating mutations, your DNA sequence is fixed and doesn’t change as you age. The epigenome does. That’s the layer of biochemical tags, including DNA methylation, that controls which genes are active as a result of your lifestyle choices and environment, and it shifts in remarkably predictable ways over a lifetime.
Researchers learned to read those shifts. The earliest, first-generation epigenetic clocks, from 2011 and 2013, showed that methylation patterns could predict chronological age.1-3 The next generation went much, much further. Rather than estimating how many birthdays you’d had, clocks like DNAm PhenoAge and GrimAge estimated how worn your biology was, and they tracked closely with the risk of age-related disease and death.4,5 A more recent measure, DunedinPACE, estimates the rate at which you’re epigenetically aging right now, in epigenetic years gained per calendar year.6
The recurring finding across all of them was striking. These epigenetic measures often predicted health outcomes, things like heart disease, cancer, frailty, and cognitive decline, as well as or better than the traditional risk factors they were measured against.5 Compared to first-generation clocks, these next-generation models were much more strongly associated with age-related outcomes, including all-cause mortality risk, in longitudinal data.
The real lesson buried in the clock
Sit with what that implies. An epigenetic clock doesn’t “know” what aging is. It learned to read aging from patterns in methylation data. And if a single, relatively simple model can pull something as profound as aging out of those signals, the signals themselves must be carrying a staggering amount of health information.
That’s the pivot. Epigenetic age was never the limit of what the epigenome could tell us. It was the first thing we figured out how to ask. The same data behind a clock can, in principle, answer far more pointed questions: which specific conditions a person is most likely to face, and how soon.
Same data, harder questions
Getting from aging to disease risk comes down to two things: a richer read of the signal and a more capable way to interpret it.
The early clocks were built on a handful, tens, or hundreds of methylation sites. A few drops of blood can now be read for about a million epigenetic signals. And where those early clocks ran on fairly simple linear models, today’s biological AI models can learn the dense, tangled patterns across all those signals that correspond to particular conditions.
Put the two together and the question itself changes. Instead of compressing the epigenome down to one number about aging, you can ask it many targeted questions at once: about metabolic health, cardiovascular risk, liver health, and lung health. The aging clock proved the signal was there. Better reads and better models are what turn that signal into specific answers.
What that looks like in practice
This is the direction the field is heading, and the direction the Infinite Epigenetics platform is built around. The same kind of methylation data that gave us indicators of biological age now sits underneath models aimed at specific chronic diseases. The company’s four lead programs target type 2 diabetes, cardiovascular disease, metabolic dysfunction-associated steatotic liver disease (fatty liver), and chronic obstructive pulmonary disease, all read from a single sample.
Contextualizing epigenetic measures
Epigenetic measures are powerful, but they’re best understood as correlational instruments. They’re strongest at the population level and most useful tracked over time rather than read as a single verdict. A risk estimate is a prompt to look closer and talk to a clinician, not a diagnosis. The science is moving fast and the validation work continues. What isn’t in question is the trajectory. The epigenome has kept turning out to hold more than we’d yet learned to read.
The epigenetic clock was a landmark, but its deepest contribution may have been the proof of concept. It showed that the body’s most dynamic data layer could be read, and that even a rough reading carried profound information about health. We’re now reading that layer at far higher resolution, with AI able to find the patterns that matter. Biological age indicators was the first question. Disease risk, earlier and more specific and more personal, is what comes next.
Infinite Epigenetics builds on more than a decade of epigenetic clock research, applying a biological AI foundation model to one of the world’s largest methylation datasets. Written for general education; not medical or investment advice.
Sources
- Bocklandt S, “Epigenetic predictor of age,” PLoS One, 2011.
- Horvath S, “DNA Methylation Age of Human Tissues and Cell Types,” Genome Biology, 2013
- Hannum et al., “Genome-wide Methylation Profiles Reveal Quantitative Views of Human Aging Rates,” Molecular Cell, 2013.
- Levine et al., “An Epigenetic Biomarker of Aging for Lifespan and Healthspan,” Aging, 2018
- Lu et al., “DNA Methylation GrimAge Strongly Predicts Lifespan and Healthspan,” Aging, 2019.
- Belsky et al., “DunedinPACE, a DNA Methylation Biomarker of the Pace of Aging,” eLife, 2022.