Barcelona scientists develop AI to detect blood stem cell aging

A scientific team in Barcelona has developed ChromAgeNet, an artificial intelligence tool designed to detect the aging of blood stem cells by analyzing the three-dimensional structure of their DNA. The research, published in the journal ‘Aging Cell’, was led by Dr. Maria Carolina Florian of IDIBELL and Dr. Paula Petrone of BSC-CNS/ISGlobal.

By utilizing convolutional neural networks applied to 3D images of mouse stem cell nuclei, the tool can identify subtle structural variations in chromatin that are imperceptible to the human eye. ChromAgeNet distinguishes between cells from young and aged mice by observing how chromatin organization changes; in younger cells, chromatin is compact and organized at the nuclear periphery, whereas it becomes more scattered in older cells.

Notably, the tool utilizes Explainable AI (XAI), addressing a common concern in biomedicine regarding the transparency of neural network decision-making. It achieved an average correct classification rate of approximately 68% using standard, low-cost DAPI staining. The researchers demonstrated the tool's potential by using it to evaluate how certain epigenetic drugs can induce chromatin changes that mimic the patterns seen in younger cells.