A machine learning model can predict anastomotic leakage within 24 hours post-minimally invasive colorectal surgery, enhancing patient selection and safety.

• Incidence of anastomotic leakage in the study was 9.8%.
• The model achieved an AUC of 0.819 and a negative predictive value of 97.9%.

This reliable early warning system allows for safer discharges, even with a 70% drop in complication rates.

• Postoperative C-reactive protein and systemic inflammation response index were strong predictors.