Construction and validation of a nomogram model for in-hospital mortality in Critically ill patients with cardiac thrombus: a retrospective study

Accurate and efficient evaluation plays an important role in improving the unfavorable short-term prognosis of critically ill patients with cardiac thrombus. However, current models have not focused on patients in the intensive care unit (ICU) and type of thrombus mainly studied was left ventricular thrombus. We extracted data on patients with cardiac thrombus admitted to the ICU from the MIMIC-IV database and the eICU-CRD. Clinical data from patients with cardiac thrombus in the MIMIC-IV database were included as a training cohort, and from the eICU-CRD were used as an external validation cohort. Based on LASSO regression and multifactor logistic regression analyses, predictors were selected and a nomogram model was developed. ROC curves, calibration pictures, and decision curve analysis were performed to assess the different performance of the nomogram model. Our finding indicated that the use of vasopressin, the acute physiology score (APS) and the use of oral anti-coagulant (OAC), as hospital mortality predictors were crucial in patients with cardiac thrombus in ICU setting ( P < 0.05). Our nomogram model showed the AUC was 0.891 (0.840–0.943) in the training set and 0.796 in the external validation set. Calibration plots demonstrated predictive consistency, and decision curves showed favorable clinical applicability. Our model demonstrates good performance in estimating the risk of in-hospital mortality among critically ill patients with cardiac thrombus, providing clinicians with a simple and dependable tool for guiding clinical treatment decisions.