Predicting coronary artery lesions in Kawasaki disease: a nomogram based on LASSO regression feature selection

BACKGROUND: Coronary artery lesions (CALs) represent the most serious complications of Kawasaki disease (KD), and the early identification of children at high risk for CALs is critical for guiding treatment strategies. This study aimed to develop a nomogram-based model to predict the risk of CALs in children with KD. METHODS: This retrospective analysis enrolled 255 children diagnosed with KD between January 2024 and May 2025. Feature selection was performed using the least absolute shrinkage and selection operator regression. Variables with non-zero coefficients were subsequently incorporated into multivariable logistic regression to identify risk factors associated with CALs. A nomogram was constructed based on these predictors, and internal validation was performed using the bootstrap resampling method. The performance of the model was assessed using receiver operating characteristic (ROC) curves, calibration curves, the Hosmer-Lemeshow goodness-of-fit test, and decision curve analysis (DCA). RESULTS: The incidence of CALs in children with KD was 29.4% (75/255). Our results identified age, clinical type, platelet count (PLT), and thrombin time (TT) as independent risk factors for CALs in children with KD. The area under the ROC curve of the model was 0.807 (95% confidence interval [CI] = 0.750-0.860). The calibration curves and the Hosmer-Lemeshow goodness-of-fit test revealed good consistency between the predicted probabilities and observed outcomes, and DCA confirmed that the model had good clinical utility. CONCLUSIONS: We developed and internally validated a nomogram-based predictive model based on age, clinical type, PLT, and TT that may facilitate the early identification of children with KD at high risk of developing CALs.