Fig. 2 | Scientific Reports

Fig. 2

From: Machine learning model for early prediction of acute kidney injury in heatstroke patients based on the first 24 h hospitalization data

Fig. 2

ROC curves and correlation analysis of candidate predictors. (A) Receiver operating characteristic (ROC) curves for individual laboratory indicators in predicting acute kidney injury (AKI). The corresponding area under the curve (AUC) values and 95% confidence intervals are indicated for Hematocrit (HCT), Serum Creatinine (Scr), Blood Urea Nitrogen (BUN), Myoglobin (Mb), Cardiac Troponin T (TnT), Interleukin-6 (IL-6) and D-Dimer. (B) Heatmap of Pearson correlation coefficients among candidate predictor variables. The color gradient reflects the strength and direction of pairwise correlations, ranging from − 1 (strong negative correlation) to + 1 (strong positive correlation). This visualization complements the variance inflation factor (VIF) analysis by illustrating potential collinearity between variables.

Back to article page