Fig 1: Construction and validation of the logistic regression prediction model. (A) ROC curve of the multivariate logistic regression prediction model, with an AUC of 0.91. (B) Nomogram generated based on the logistic regression prediction model. (C) Calibration curve showing good performance of the model on the training set. However, bias-corrected results indicate some calibration deviations in the mid-to-high probability range. The HL test (P=0.4783) suggests no significant calibration issues. (D) Decision curve analysis (DCA). The x-axis represents the predicted probability threshold, and the y-axis indicates net benefit, demonstrating the clinical utility of the model across varying thresholds. (E) ROC curve for univariate analysis of FCGR1A, showing a predictive performance with an AUC of 0.772. (F) ROC curve for univariate analysis of CBS, showing a predictive performance with an AUC of 0.796.