Fig. 6
Construction and validation of diagnostic models for Type 2 Diabetes (T2D) and Sjögren’s syndrome (SS). (a) T2D diagnostic nomogram (GSE18732). Predicts T2D risk based on two biomarkers. Points scale (top axis): Score for each gene expression level; Total points (middle axis): Summed scores; Risk probability (bottom axis): Predicted T2D probability. Gray vertical lines: Projection path for sample calculation. (b) Decision curve analysis (DCA) for T2D model. X-axis: Threshold probability; Y-axis: Net benefit; Darkslateblue curve: Diagnostic model; Mistyrose line: “Treat all” strategy; Darksalmon line: “Treat none” strategy. (c) and (d) ROC curves for T2D validation: (c) T2D dataset (GSE18732) evaluation: X-axis: 1—Specificity; Y-axis: Sensitivity; Blue curve: Diagnostic model (AUC = 0.705, 95% CI 0.597–0.813), (d) Independent T2D cohort (GSE29221) validation: Components identical to (c) (AUC = 0.764, 95% CI 0.562–0.966), (e) SS diagnostic nomogram (GSE40611). Configuration identical to a with SS risk prediction. Pink vertical lines: Sample calculation path. (f) DCA for SS model. Component scheme identical to (b). (g) and (h) ROC curves for SS validation: (g) SS dataset (GSE40611) evaluation: Red curve: Diagnostic model (AUC = 0.807.X, 95% CI 0.657–0.958), (h) Independent SS cohort (GSE154926) validation: Components identical to g (AUC = 0.738, 95% CI 0.540–0.936).
