Machine learning approaches to predict 30-day mortality following percutaneous coronary intervention in an Australian population

BackgroundPre-procedural risk prediction of 30-day all-cause mortality after percutaneous coronary intervention (PCI) aids in clinical decision-making and benchmarking hospital performance. This study aimed to identify pre-procedural factors to predict the risk of 30-day all-cause mortality post-PCI using machine learning (ML) approaches. MethodsThe study analysed 93,055 consecutive PCI procedures. Boruta feature selection method was used to identify key predictive variables. Seven ML algorithms were employed for models development and validation. Model performance was assessed using standard metrics for validation dataset. SHapley Additive exPlanations (SHAP) method was used to explain leading predictive variables. ResultsAmong the seven ML algorithms, the Extreme Gradient Booster (XGB) had the better performance across most metrics, such as accuracy (86.7%), root mean square error (36.5%), specificity (82.5%), precision (54.0%), F1 score (52.7%), and Brier score (13.3%). The XGB model also demonstrated strong discriminatory power, achieving a receiver operating characteristics-area under the curve (ROC-AUC) of 85.5% (95% CI: 83.5%-87.4%). The XGB model identified left ventricular ejection fraction (LVEF), acute coronary syndrome (ACS), estimated glomerular filtration rate (eGFR), age, and complex lesion as the five leading factors associated with 30-day mortality post-PCI. Other factors, in order, were cardiogenic shock, body mass index (BMI), intubated out-of-hospital cardiac arrest (OHCA), lesion location, mechanical ventricular support, gender, and peripheral vascular disease (PVD). ConclusionThe XGB algorithm was identified as the best predictive model for 30-day all-cause mortality post-PCI. It is essential to underscore the need for further validation of the model with external data to ensure its applicability to other populations. WHAT IS ALREADY KNOWN ON THIS TOPICO_LIrisk-adjustment model for an Australian percutaneous coronary intervention (PCI) patient population was previously developed to predict 30-day mortality post-PCI using traditional regression model. C_LIO_LIknowledge, patient characteristics, and clinical practices evolve over time, requiring frequent model updates to reflect new evidence, guidelines, and interventions C_LI WHAT THIS STUDY ADDSO_LIA machine learning (ML)-based preprocedural risk prediction model for 30-day mortality post-PCI was developed. The Extreme Gradient Booster (XGB) model was identified as the top performer in predicting 30-day all-cause mortality post-PCI. The model selected left ventricular ejection fraction, acute coronary syndrome, estimated glomerular filtration rate, age, and complex lesion as the top influential factors. C_LI HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYO_LIRisk prediction models aid clinical decision-making, enhance patient counselling, improve care quality, inform healthcare policies, and advance research. C_LI