A new machine learning model predicts anastomotic leakage risk post-gastrectomy, crucial for improving outcomes.

• The model shows an AUC of 0.871 with a sensitivity of 71.2% and specificity of 87.3%.
• Using CRP levels within three days post-surgery as a key predictor can boost negative predictive value to 98.9% at a higher sensitivity threshold.

Surgeons can use this model to identify at-risk patients early, guiding intervention strategies and potentially reducing complications.

• Continued multicenter validation is needed for broader clinical application.