DNA Changes Track Human Age with Remarkable Precision

Time stamps itself in our cells through distinct patterns of DNA changes, allowing scientists to determine a person’s age with unprecedented accuracy using just a small blood sample.

Researchers at the Hebrew University of Jerusalem have developed a groundbreaking method that can determine a person’s chronological age with a median error of just 1.36 years for people under 50. Their technique, called MAgeNet, uses artificial intelligence to analyze how DNA methylation—chemical tags added to our genetic material—changes in predictable patterns as we age.

“It turns out that the passage of time leaves measurable marks on our DNA,” explains Professor Tommy Kaplan, one of the study’s senior authors. “Our model decodes those marks with astonishing precision.”

Reading time’s fingerprints at the molecular level

The research team, led by Bracha Ochana and Daniel Nudelman, discovered that by focusing on just two specific regions of the genome, they could achieve far greater accuracy than previous methods. What makes their approach unique is the analysis of methylation patterns at the level of individual DNA molecules rather than averaging across many cells.

Using deep bisulfite sequencing, which reveals the methylation status of multiple adjacent DNA sites, the team found two distinct ways our cells record the passage of time:

• Some DNA regions accumulate methylation changes randomly and independently at each site, like individual grains of sand falling through an hourglass
• Other regions change in coordinated “blocks,” with multiple adjacent sites switching their methylation status together
• These patterns create a molecular timestamp that becomes more pronounced with each passing year

“We found that some age-related methylation changes could originate from cell-intrinsic changes within specific cell types, while others reflect shifts in blood cell composition over time,” notes the research team. This discovery helps explain how our bodies track the passage of time at the cellular level.

Accuracy that outperforms all previous methods

The researchers tested their method on blood samples from more than 300 healthy individuals aged 17-78, achieving remarkable precision. For individuals under 35, the median error was just 0.9 years—far more accurate than any previous epigenetic clock.

What’s particularly striking is the method’s consistency. Unlike other biological markers that can be influenced by lifestyle or environment, MAgeNet’s age predictions remained steady regardless of factors like smoking status, body mass index, or sex. The researchers confirmed this by analyzing samples from the Jerusalem Perinatal Study, which tracked individuals over a 10-year period.

When the team examined how their predictions changed over time, they found that if a person’s predicted age deviated from their actual age at the first measurement, that same deviation persisted a decade later. This suggests that early life events might permanently shift a person’s “epigenetic age,” after which the passage of time is faithfully recorded.

Applications from forensics to medical research

Perhaps most remarkable is how little DNA the method requires. The researchers found they could accurately predict age from as few as 50 DNA molecules—equivalent to the genetic material from just a handful of cells. This has significant implications for forensic science, where investigators often have access to only trace amounts of biological material.

“The ability to determine age from such a small sample could transform how we approach criminal investigations,” says Professor Ruth Shemer, another senior author on the study.

The technique also has potential applications in aging research and personalized medicine. By distinguishing between chronological age (time since birth) and biological age (the body’s physiological state), doctors could potentially tailor treatments to an individual’s unique aging profile.

A new understanding of how we age

Beyond its practical applications, the research provides fundamental insights into how our bodies measure time. The team discovered that age-dependent methylation changes occur regionally across clusters of CpG sites (locations where methylation occurs) either stochastically or in a coordinated block-like manner.

This mechanism appears to be a universal feature of human aging, as the researchers demonstrated by successfully applying their model to urine samples (with a median error of 2.45 years) and, to a lesser extent, saliva samples.

The study, published in Cell Reports, represents a significant advance in our understanding of the biology of aging and offers a powerful new tool for both research and practical applications. As Professor Yuval Dor summarizes: “It’s a powerful example of what happens when biology meets AI.”