Leveraging advanced deep learning architectures to scrutinize DNA methylation profiles, researchers at the Hebrew University have demonstrated the capacity to predict chronological age, defined as elapsed time since birth, with a median precision of 1.36 to 1.7 years for individuals below the age of 50. These groundbreaking findings have been disseminated in the esteemed journal Cell Reports.
Utilizing ultra-deep sequencing of more than 300 blood specimens from healthy volunteers, our investigation reveals that age-correlated methylation alterations manifest regionally across aggregations of CpG sites, proceeding either randomly or in a synchronized, block-like fashion. Attribution for this visual representation is extended to Ochana et al., with the associated digital object identifier being 10.1016/j.celrep.2025.115958.
“It has been observed that the progression of time inscribes discernible imprints upon our genetic material,” commented Professor Tommy Kaplan of the Hebrew University.
“Our computational model possesses the capability to decipher these biological markers with remarkable accuracy.”
“The fundamental principle lies in the temporal evolution of our DNA through a biological phenomenon known as methylation, which involves the chemical ‘labeling’ of DNA by methyl groups (CH3).”
“By concentrating our analysis on merely two critical domains within the human genome, our research cohort successfully interpreted these molecular transformations at an individual molecule level, subsequently employing deep learning techniques to translate these insights into precise age estimations.”
Within the scope of this investigation, Professor Kaplan and his associates meticulously examined blood samples collected from over 300 healthy individuals, in addition to data derived from a decadal-long longitudinal study originating from the Jerusalem Perinatal Study.
The predictive model developed by the research team exhibited consistent performance across a spectrum of variables, including but not limited to lifestyle factors such as smoking habits, body mass index, gender, and even discernible indicators of physiological aging.
Beyond its potential applications in the medical field, this innovative methodology holds the promise of revolutionizing forensic science, empowering experts to ascertain the age of a person of interest from even minute quantities of DNA—a feat that current analytical tools find challenging to accomplish.
“This advancement provides us with an unprecedented perspective on the intricate mechanisms of aging at the cellular level,” stated Professor Yuval Dor from the Hebrew University.
“It stands as a compelling illustration of the synergistic outcomes achievable when biological inquiry converges with artificial intelligence.”
The researchers have identified novel patterns in the manner in which DNA undergoes modifications over time, suggesting that cellular processes may encode age through both stochastic events and coordinated, episodic changes, akin to the operation of internal biological chronometers.
“The significance extends beyond merely ascertaining an individual’s chronological age,” articulated Professor Ruth Shemer, also affiliated with the Hebrew University.
“It delves into understanding how cellular structures meticulously track the passage of time, on a molecule-by-molecule basis.”
“This line of inquiry possesses the potential to fundamentally reshape our future approaches to health, the aging process, and the very concept of identity,” the scientists concluded.
“From facilitating physicians in personalizing therapeutic interventions based on an individual’s actual biological timeline to furnishing forensic investigators with a potent new instrument for resolving criminal cases, the capacity to directly ascertain age from DNA heralds the advent of significant advancements across the scientific, medical, and legal disciplines.
“Furthermore, it deepens our comprehension of the aging process, bringing us a definitive step closer to unraveling the intricacies of the body’s intrinsic temporal regulation system.”
_____
Bracha-Lea Ochana et al. Time is encoded by methylation changes at clustered CpG sites. Cell Reports, published online July 14, 2025; doi: 10.1016/j.celrep.2025.115958
