An AI system trained on microscope images of human tissue can estimate how old an organ is biologically, and the answer often differs from the age on a person's birth certificate. Researchers at the CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences and the Ludwig Boltzmann Institute for Network Medicine at the University of Vienna analyzed more than 25,000 tissue samples spanning 40 tissue types and found that organs within the same body age at different rates. The work was published in Nature Medicine.
Most efforts to measure biological age have relied on molecular markers such as DNA methylation patterns or gene expression. The team led by André Rendeiro, with co-first authors Ernesto Abila, Iva Buljan and Yimin Zheng, took a different route: they looked at how the physical architecture of tissue changes over time. Using samples from 983 donors collected by the Genotype-Tissue Expression Project, covering brain, heart, lung, pancreas, skin and intestine among others, they converted tissue slices into high-resolution digital images and fed roughly 480 million image tiles through modern computer vision models.
The practical appeal is that these age signals turned up not only in solid organs but could also be detected from blood samples, which are far easier to obtain than a biopsy of the brain or pancreas. If an organ's structural age can be read off routinely collected material, it opens a path toward tracking how specific organs are faring long before symptoms of disease appear - the kind of organ-specific monitoring that current whole-body age estimates cannot provide.
The authors present the tissue clocks as a framework for studying aging rather than a finished clinical tool, and note the potential for disease monitoring and early diagnosis as avenues to explore. Whether reading a tissue's biological age can predict who goes on to develop a particular disease, and how early that signal appears, remains for future work.