Phoenix 24
The biological clock may leave patterns that artificial intelligence can learn to recognize.
Barcelona, Spain
A research team in Barcelona has developed an artificial intelligence system capable of distinguishing young from aged blood stem cells by analyzing the three-dimensional organization of DNA inside the cell nucleus. The tool, called ChromAgeNet, uses deep learning to identify structural changes in chromatin that are difficult for the human eye to detect.
The research was led by Maria Carolina Florian of the Bellvitge Biomedical Research Institute and Paula Petrone of the Barcelona Supercomputing Center and ISGlobal. Their work suggests that aging leaves measurable signatures in the architecture of hematopoietic stem cells, the cells responsible for generating blood and supporting immune function.
To train the system, the researchers used convolutional neural networks on 3D images of stem-cell nuclei from mice. ChromAgeNet correctly classified roughly 68 percent of the cells as young or aged using DAPI, a standard and relatively inexpensive staining technique. The system is also designed as explainable AI, allowing scientists to examine which parts of the images influenced its predictions.
One of the clearest patterns appears at the edge of the nucleus. In younger cells, chromatin tends to remain more compact and organized near the nuclear periphery. In aged cells, that structure becomes more dispersed, suggesting that biological aging may be visible not only through genetic or molecular markers, but also through the physical organization of DNA.
The team also tested whether the model could detect changes after aged cells were exposed to epigenetic drugs. Some treatments produced chromatin patterns that appeared more similar to those found in younger cells, suggesting that ChromAgeNet could eventually help researchers screen compounds with potential rejuvenating effects.
The implications are promising but still preliminary. The current system is a proof of concept developed with mouse hematopoietic stem cells, and the researchers are now working to adapt it for human cells. Any clinical application would require extensive validation before the technology could influence treatment decisions or drug development.
The project is particularly significant because it combines biomedical research, supercomputing and explainable artificial intelligence. The team has also released the code and a public database of 3D stem-cell images, making the work available to researchers who want to test, refine or extend the method.
The larger objective is not simply to extend life, but to understand whether aging tissues can remain functional for longer. If tools like ChromAgeNet eventually help identify treatments that preserve blood production and immune resilience, artificial intelligence could become part of a broader effort to increase healthy life expectancy rather than merely longevity.
The future of anti-aging research may begin with learning how to read the architecture of a single cell.