An innovative algorithm from routine data can effectively classify 30-day postoperative complications in hepatobiliary surgery, helping surgeons better understand patient outcomes.

• Among 959 liver resections, the algorithm achieved an impressive macro-f1-score of 0.962 and a sensitivity of 0.950.
• Major complications were observed in 18% of cases, with a 2.6% mortality rate.

This tool outperforms machine learning approaches and allows for real-time complication surveillance and quality evaluation.

• Misclassification was low at 3.2%, primarily linked to ICU coding issues.