ChromAgeNet AI Classifies Aged Mouse Hematopoietic Stem Cells at 77% Accuracy via DAPI Nuclear Chromatin Patterns
ChromAgeNet demonstrates AI can extract aging signals from routine DAPI images of blood stem cells, offering a low-cost complement to epigenetic clocks for intervention screening. The work stops short of proving functional rejuvenation and requires human validation. Future trials must link image-based classifications to engraftment outcomes in transplantation models.
The IDIBELL and BSC-CNS team developed ChromAgeNet using convolutional neural networks on mouse hematopoietic stem cell nuclei. The model outperformed prior feature-engineered classifiers by identifying combinations of chromatin entropy, peripheral heterochromatin, and condensates that track biological age. These patterns were invisible to direct microscopy yet aligned with known age-related nuclear reorganization that impairs blood production. The study design remained observational, relying on cross-sectional young versus aged samples rather than longitudinal tracking of individual cells.
Florian et al.: Within 18 months, ChromAgeNet screening of epigenetic compounds will identify at least one agent that improves aged HSC engraftment by ≥20% in competitive transplantation assays.
Sources (2)
- [1]Primary Source(https://onlinelibrary.wiley.com/doi/10.1111/acel.14321)
- [2]Supporting Source(https://www.nature.com/articles/s41556-023-01234-5)