Four-dimensional left atrial quantification for predicting major adverse cardiovascular events in patients with uremia: an interpretable LASSO-logistic model

Background This study aimed to develop a machine learning model using four-dimensional left atrial quantification (4D-LAQ) parameters to the risk of major adverse cardiovascular events (MACE) in patients with uremia. Methods A total of 223 patients with uremia treated at the Second Affiliated Hospital of Hainan Medical University between January 2021 and January 2026 were retrospectively enrolled and followed up for the occurrence of MACE. LASSO regression was utilized for feature selection, followed by the development of eight prediction models. Logistic regression (LR) was identified as the final algorithm based on DeLong’s test. A nomogram was constructed, and Shapley Additive Explanations (SHAP) analysis was performed to enhance model interpretability. Model performance was comprehensively evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA). Intra- and inter-observer reproducibility were assessed using the intraclass correlation coefficient (ICC). Results During the follow-up period, 94 patients (42.1%) developed MACE. The MACE group exhibited left ventricular remodeling, mainly characterized by left atrial volume enlargement and a decrease in strain. The LR model achieved an AUC of 0.840 (95% CI: 0.749–0.931) in the validation set. Multiple validation approaches, including five-fold cross-validation, nested 5×5 cross-validation, and bootstrap internal validation, yielded AUCs of 0.873, 0.856, and 0.869, respectively. The calibration analysis showed good consistency (Hosmer-Lemeshow test, P > 0.05). The AUC of DCA (AUDC) was 0.208 (0.206 - 0.208), and the maximum net benefit was 0.43. The optimal risk stratification threshold, determined by the maximum Youden index in the training set, was 0.41. SHAP analysis identified left atrial reservoir strain (LASr) as the most influential predictive feature in this cohort. Conclusion The LASSO-LR model integrating 4D-LAQ parameters may help MACE risk in patients with uremia, with LASr emerging as an important predictor.