Interpretable machine learning model for new-onset atrial fibrillation prediction in critically ill patients: a multi-center study

Topics

Incident atrial fibrillation · Critically ill patients · MIMIC-IV database · MIMIC-III database · LASSO regression · Feature selection · XGBoost · SHapley Additive exPlanations (SHAP) · Machine learning prediction model · external validation · Area under the curve · Model calibration · Clinical decision support interface · Mechanical ventilation · Urine output · Sepsis · Blood urea nitrogen · Percutaneous arterial oxygen saturation · Continuous renal replacement therapy · Risk stratification threshold