Development and internal-external validation of a nomogram model for lower extremity deep vein thrombosis in patients after ruptured intracranial aneurysm surgery based on dynamic D-dimer trajectory.

Abstract

This study aimed to develop and validate a nomogram model for predicting lower extremity deep venous thrombosis (LEDVT) after ruptured intracranial aneurysm (RIA) surgery based on dynamic D-dimer trajectory, to support accurate identification of high-risk populations and individualized thrombosis prevention in clinical practice. A multicenter retrospective cohort study was conducted, enrolling consecutive RIA patients undergoing emergency surgery at two hospitals from January 2021 to August 2024. Patients were split into training and test sets at an 8:2 ratio; an independent external validation set included patients from another hospital. D-dimer trajectory analysis was performed via Gaussian Mixture Model (GMM), LASSO regression screened predictive variables, and multivariate Logistic regression was used to construct the nomogram. Model performance was assessed by AUC, calibration curves, Hosmer-Lemeshow (H-L) test and DCA. Overall, 488 eligible patients were included (training set n = 294, test set n = 74, external validation set n = 120), with a mean age of 60.1 ± 12.0 years and 65.2% female. Four D-dimer trajectory subgroups were identified by GMM. LASSO regression screened age, surgical type, GCS score and D-dimer trajectory as key predictors. The model showed favorable discrimination (training set AUC = 0.76, test set = 0.81, external validation = 0.75), good calibration (all H-L P > 0.05) and strong clinical net benefit via DCA. This validated nomogram integrating four core factors can accurately predict postoperative LEDVT risk in RIA patients.