A Nomogram for Predicting 28-Day Mortality inCritically Ill Patients With Prostate Cancer: ARetrospective Cohort Study Based on the MIMIC-IVDatabase

Background: Despite their clinical complexity, patients with prostate cancer (PCa) requiring intensive care remain an understudied demographic for which precision prognostic instruments are lacking. We sought to establish and verify a robust nomogram tailored to predict the 28-day mortality risk for this specific cohort. Methods: We utilised retrospective data from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database. Patients were randomly assigned to training and validation sets (7:3) by using random seeds. Least Absolute Shrinkage and Selection Operator (LASSO) regression with 20-fold cross-validation and the 1SE rule was employed to select predictors and construct the model. Model performance was evaluated by employing receiver operating characteristic (ROC) curves, calibration plots with Hosmer-Lemeshow tests and decision curve analysis (DCA). Results: The analysis of 529 patients (370 training/159 validation) yielded six decisive predictors. Our proposed nomogram exhibited markedly higher discriminative accuracy than traditional metrics. Specifically, it achieved an area under the receiver operating characteristic curve (AUROC) of 0.814 in the training set, eclipsing Simplified Acute Physiology Score II (SAPS II) (0.737), Sequential Organ Failure Assessment (SOFA) (0.645) and the Charlson Comorbidity Index (CCI) (0.734). These findings were mirrored in the validation phase (AUROC: 0.809). On the validation set, we also found that the AUROC of our model was 0.809, whereas those of SAPS II, SOFA and CCI were 0.764, 0.686 and 0.725, respectively. High p values in Hosmer-Lemeshow tests (p > 0.05) reflected strong calibration. Meanwhile, DCA curves underscored our nomogram’s superior net clinical benefit over conventional scoring systems. Conclusions: Our established 28-day mortality prediction model for critically ill patients with PCa facilitates early risk stratification and intervention, potentially improving survival outcomes in this high-risk population.