Researchers from IDIBELL, the Barcelona Supercomputing Center, and ISGlobal have developed ChromAgeNet, an AI-powered tool that detects aging-related changes in hematopoietic stem cells (HSCs) — the blood-producing cells that decline in function as we age. Published in Aging Cell, the tool uses a convolutional neural network to analyze 3D microscopy images of cell nuclei stained with DAPI, a simple and widely available DNA dye, identifying subtle shifts in chromatin organization that the human eye simply can't catch.

The model doesn't just classify cells — it also reveals which structural features matter most for aging, including chromatin entropy, peripheral heterochromatin, and specific chromatin condensates. Researchers also tested ChromAgeNet as a screening tool for epigenetic drugs, showing it can detect chromatin changes consistent with a "younger" state after treatment.

Key Takeaways:

• ChromAgeNet achieved 77% accuracy in distinguishing young from aged HSCs, outperforming prior feature-based machine learning models
• Key predictive features include chromatin entropy, peripheral heterochromatin, and chromatin condensates
• The tool is compatible with high-throughput microscopy, enabling large-scale compound screening for rejuvenation strategies
• The team has made their 3D HSC imaging dataset publicly available to support future research

Why it matters: As the population ages, understanding how blood stem cells deteriorate is critical for developing therapies for age-related blood disorders. ChromAgeNet offers a scalable, low-cost way to study cellular aging and fast-track the search for treatments that could restore stem cell function.