Fig. 2: Overview of the study framework.

a Data extraction and pre-processing steps performed in this study. b The proposed end-to-end model automatically performs real-time imputation of missing values and predicts patient mortality risk each time a new measurement is obtained. c Compared with traditional imputation methods, our approach more effectively captures temporal dependencies and interactions among features. d The modelling performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, decision curve analysis (DCA), and temporal alert efficiency curve. e The SHapley Additive exPlanations algorithm was used to interpret the real-time individualized prediction results after missing values were imputed, and the SHAP risk scores were visualized. eICU-CRD eICU Collaborative Research Database, MIMIC Medical Information Mart for Intensive Care, SICdb Salzburg Intensive Care Database, AUC area under the receiver operating characteristic curve, DCA decision curve analysis, HR heart rate, RR respiratory rat, SBP systolic blood pressure, ALT alanine aminotransferase, WBC white blood cell. Some elements in the figure were drawn by Figdraw.