The Prognostic Role of Hematological Markers in Acute Pulmonary Embolism: Enhancing Risk Stratification

Figures

• 

  Figure 1. Study flow diagram. PE, pulmonary embolism.

• 

  Figure 2. ROC analysis for NLR as a predictor of in-hospital adverse events. The solid red line represents the ROC curve for NLR, with an area under the curve (AUC) of 0.712 (p < 0.001), indicating fair discriminative ability. The dashed blue lines represent the 95% confidence interval (CI) for the ROC curve. The diagonal grey line indicates the line of no discrimination (AUC = 0.5).

• 

  Figure 3. ROC curves for (a) sPESI as a predictor of adverse in-hospital events and (b) a combined parameters sPESI and NLR as a predictor of adverse in-hospital events. The solid line in each panel represents the ROC curve derived from the predictive model indicated. The outer dotted blue lines depict the 95% confidence interval (CI) for the area under the curve (AUC), reflecting the precision of the estimate. The diagonal line represents the line of no discrimination (AUC = 0.5). In (a), the blue ROC curve illustrates the predictive performance of sPESI alone (AUC = 0.805). In (b), the red ROC curve represents the combined model of sPESI and NLR, which shows an improved AUC of 0.844.

• 

  Figure 4. ROC curves for (a) PESI as a predictor of adverse in-hospital events and (b) a combined parameters PESI and NLR as a predictor of adverse in-hospital events. In both panels, the main ROC curve (solid stepped line) illustrates the relationship between sensitivity and 1-specificity for each model. The red diagonal line represents the line of no discrimination (AUC = 0.5), serving as a reference for random performance. The outer thin blue lines represent the 95% confidence interval (CI) bands, indicating the precision around the ROC estimate. The red squares mark the actual performance points used to construct the ROC curve. In panel (a), the PESI score alone demonstrates an AUC of 0.822, with sensitivity and specificity of 71.34% and 79.29%, respectively, at a cutoff >115. In panel (b), the combined model of PESI and NLR shows an improved AUC of 0.839, with higher sensitivity (79.14%) and slightly reduced specificity (74.83%), indicating a statistically significant improvement in predictive performance (p = 0.005).

• 

  Figure 5. ROC curves for (a) ESC risk class as a predictor of adverse in-hospital events and (b) a combined parameters ESC risk class and NLR as a predictor of adverse in-hospital events. In panel (a), the ESC risk class yielded an AUC of 0.880 (95% CI: 0.857–0.901; p < 0.001), with a sensitivity of 73.17% and specificity of 98.64%. In panel (b), the combined ESC risk class and NLR model showed improved discrimination with an AUC of 0.921 (95% CI: 0.901–0.938; p < 0.001), sensitivity of 77.91%, and specificity of 94.77%. In both panels, the solid ROC curve represents the model performance, the surrounding dotted blue lines indicate the 95% confidence intervals, and the diagonal dashed line represents the line of no discrimination (AUC = 0.5). The improvement in predictive accuracy with the combined model was statistically significant (p < 0.001).