Development and Validation of an AI-Based Tool for Identifying Responders and Stratifying Risks in Ulinastatin Treatment in Cardiac Surgery: A Multicenter Real-world Study

This is a multicenter, retrospective, real-world observational study aimed at developing and validating an artificial intelligence-based tool for identifying ulinastatin treatment responders and risk stratification in cardiac surgery patients undergoing cardiopulmonary bypass (CPB).

Ulinastatin, a glycoprotein extracted from human urine, has shown potential benefits in reducing postoperative complications and inflammatory responses in cardiac surgery. However, evidence supporting its efficacy and optimal application in specific patient populations remains insufficient.

This study will collect clinical data from approximately 4 tertiary cardiac centers in China, including patients who underwent cardiac surgery with CPB. Using machine learning algorithms (such as weighted K-modes clustering and XGBoost), the study aims to: (1) construct a multicenter real-world database for cardiac surgery; (2) identify clinical characteristics associated with ulinastatin treatment response; (3) develop and validate an AI-based risk stratification tool to assist clinical decision-making. This study may provide evidence-based guidance for personalized perioperative anti-inflammatory treatment in cardiac surgery.