A new machine learning model using endoscopic ultrasound can differentiate main-duct intraductal papillary mucinous neoplasm (MD-IPMN) from chronic pancreatitis (CP), which is crucial for surgical decision-making.

• Key features to note: intraductal nodules (2.4% CP vs 20.0% MD-IPMN, p = .003) and ductal stones (77.1% CP vs 2.5% MD-IPMN, p < .001).
• The model achieved a high accuracy (AUC 0.946) and remained effective in stone-negative cases (AUC 0.830).

This tool could streamline evaluations when imaging is unclear, potentially improving patient outcomes.