Prognostic value of the BAB index and a machine learning model integrating the BAB index for predicting mortality in...

The high mortality in ST-segment elevation myocardial infarction (STEMI) is associated not only with organ dysfunction and complications, but also with nutritional status. We aim to develop and validate a simple prognostic tool based on routinely serum biomarkers for predicting short- and long-term mortality in patients with STEMI, and to assess its contributing role in machine learning (ML) models.

Observational multicenter data from the Tianjin Coronary Artery Disease (CAD) Database (2010-2021) were analyzed. The predictive abilities of biomarkers were identified via multivariable Cox regression. The BAB Index was calculated as Log₁₀(NT-proBNP × ALT × BUN). Prognostic performance was evaluated by area under the curve (AUC) and compared with the CAMI-STEMI score. Validation included Cox regression, restricted cubic spline analysis (RCS), Kaplan-Meier survival, and subgroup analyses. ML models incorporating the BAB Index were constructed to verify the contributing roles of the BAB index in predicting 1-month and 1-year mortality.

Among 8,002 STEMI patients, BAB Index showed strong discriminatory power for 1-month (AUC = 0.804) and 1-year mortality (AUC = 0.794), comparable to the CAMI-STEMI score (P = 0.641). Higher BAB Index were independently associated with increased mortality (P < 0.001). RCS revealed a linear relationship, and Kaplan-Meier analysis confirmed worse survival with higher BAB Index (P < 0.001). Subgroup analyses demonstrated consistent findings. The XGBoost model achieved the highest performance for both 1-month (AUC 0.873) and 1-year mortality (AUC: 0.871), with BAB Index ranked among the top predictive features.

BAB Index is a simple, effective tool for predicting short- and long-term mortality in STEMI. BAB index maintains a leading position among interpretable ML models.