LongevityAI01 sources

ChromAgeNet reads ageing in the nuclei of blood stem cells

Figure from the ChromAgeNet paper: young and aged mouse blood stem cells imaged under a confocal microscope and fed to a neural network, beside grids of stained stem-cell nuclei shown in false colour.

Iañez Picazo et al., Aging Cell 2026, Figure 1CC BY

Researchers at the Bellvitge Biomedical Research Institute (IDIBELL), the Barcelona Supercomputing Center and ISGlobal announced on 28 September 2026 a paper in Aging Cell describing an AI model called ChromAgeNet that judges whether a blood stem cell is young or old from a 3D image of its nucleus.

The model is a convolutional neural network trained on images of mouse hematopoietic stem cells, the cells that renew blood throughout life, stained to show how their DNA is packed. It separated young from aged cells with 77 percent accuracy. The features it relied on include the disorder of chromatin and the layer of tightly packed DNA at the edge of the nucleus.

The team, led by Maria Carolina Florian at IDIBELL and Paula Petrone at BSC and ISGlobal, then used the model as a screen. Aged cells treated with drugs that act on chromatin or on Rho GTPase signalling shifted toward a younger score, which the group presents as a way to test rejuvenation candidates on single cells. The work is in mouse cells only, and the imaging datasets have been released publicly.

Sources

  1. [1]Artificial intelligence identifies signs of aging in blood stem cells through images of the nucleusIDIBELL··Press release