Back

AI-guided analysis of human pancreatic islet sociology reveals distinct cell compositional changes in type 1 diabetes

Ward, C.; Banks-Tibbs, T.; Thorpe, H.; Giles, A. M.; Mabry, S. J.; Liu, J.-J.; Chang, H.-C.; Yang, J.; Carney, A. F.; Shaikh, H. M.; Woolley, L. M.; Joseph, P. N.; Kozel, J. D.; Rocha, E. M.; Rutter, G. A.; Tseng, G. C.; Liu, S.; Pless, L. L.; Freyberg, Z.

2026-06-09 pathology
10.64898/2026.06.05.730081 bioRxiv
Show abstract

Human pancreatic islets exhibit greater anatomic and cellular heterogeneity than previously appreciated, raising fundamental questions about how their composition varies with age, sex, region, and islet size and how type 1 diabetes (T1D) alters these relationships. Yet these questions remained largely unresolved due to the bottleneck of manual tissue inspection. Here, we developed an integrated artificial intelligence (AI)-guided imaging, processing, and statistical pipeline enabling unbiased, high-throughput analysis of more than 2 million candidate islets from 106 non-diabetic (ND) and T1D donors. We identified age-, region-, sex-, and islet size-dependent differences in islet distribution and composition between ND and T1D donors. Profound {beta}-cell loss in T1D was accompanied by reciprocal -cell expansion, whereas {delta}-cells and pancreatic polypeptide cells were largely resilient. Cell area and pseudotime analyses uncovered regional and age-dependent trajectories of islet remodeling across T1D progression, along with distinct patterns of cytoarchitectural reorganization of the endocrine pancreas.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.