Fig. 4

The efficacy of prediction models for severe AKI with different composite inflammatory biomarkers. A In the model with MLPR alone, the AUC value was 0.853. With the progressive enrollment of variables, the AUC values increased from 0.910, 0.925 to 0.926. B The Delong test revealed that the AUC of the full model with MLPR was significantly higher than that of the full model without MLPR (AUC: 0.926 vs. 0.831, Z = 5.111, p < 0.001). C In validation cohort, the AUC of full model reached 0.968. D In the model with NLPR alone, the AUC value was 0.847. With the progressive enrollment of variables, the AUC values increased from 0.909, 0.922 to 0.923. E The Delong test revealed that the AUC of the full model with NLPR was significantly higher than that of the full model without NLPR (AUC: 0.923 vs. 0.831, Z = 5.278, P < 0.001). F In validation cohort, the AUC of full model reached 0.970