A machine-learning clock trained on healthy colon tissue estimated chronological age with a mean absolute error of 3.9 years. When researchers applied it to tissue from people with several medical conditions, the colon often appeared epigenetically older.
The Scientific Reports study adds to a shift in aging-clock research: away from one universal number and toward tools built for specific organs.
Why the colon needs its own clock
Epigenetic clocks use chemical marks on DNA, especially methylation at CpG sites, to predict age. Many widely used clocks were trained across blood or multiple tissues. That makes them broad, but it can blur tissue-specific biology.
The colon presents an extra complication. Tissue from the proximal and distal colon has distinct developmental, microbial and disease contexts. The researchers therefore trained a machine-learning model on DNA methylation data from healthy colon samples while accounting for both chronological age and anatomical location.
The final model used far fewer CpG sites and training samples than many general-purpose clocks. In the reported evaluation, predicted and chronological age were tightly correlated at 0.978, with a mean absolute error of about 3.9 years. That is prediction accuracy within the study data, not proof that the score forecasts disease or lifespan.
What accelerated age means here
Age acceleration is the gap between the clock’s predicted age and the age expected for a person of that chronological age. A positive value means the tissue’s methylation pattern looks older according to this model.
Applied to independent datasets, the clock reported higher age acceleration in colon tissue associated with HIV, inflammatory bowel disease and colonic polyps. The direction was consistent with the idea that chronic inflammation or disease can alter the tissue’s molecular aging pattern.
But “older” is a statistical description, not a microscopic diagnosis. It does not mean every cell has aged faster, and it does not show that epigenetic acceleration caused the condition. Disease, treatment, cell composition, inflammation and sample handling can all influence methylation.
The aspirin result needs restraint
The paper also reported that aspirin treatment was associated with partial deceleration of the colon clock. That is not evidence that people should take aspirin to slow biological aging.
An association can reflect who was prescribed aspirin, why they took it, dose and duration, other health differences, or features of the source dataset. Aspirin also has real bleeding risks, and preventive use depends on age and cardiovascular risk. Only a randomized trial designed around a validated clinical endpoint could establish a causal benefit.
Useful research tool, uncertain clinical meaning
Organ-specific clocks may help scientists test whether local disease processes and interventions leave a coherent aging signal. They may also be more sensitive to tissue change than a blood test.
The tradeoff is narrower validation. A clock can fit its training population well and still perform differently in other laboratories, ancestries, biopsy locations or disease states. It also needs to show that changes track outcomes people care about. A lower clock age is not automatically better health.
That issue is common across the field. Our review of which aging clocks respond to longevity interventions found that clocks can disagree even when applied to the same intervention.
Bottom line: The new clock appears technically accurate for estimating age in the studied colon samples and detected plausible disease-associated patterns. It is a research instrument, not a clinical test, and the aspirin association is a hypothesis rather than a prescription.
Primary source: Scientific Reports, August 25, 2026