Fig. 3

Defining risk groups using decision curve analysis and performance metrics across admission probability thresholds. (A) The Decision Curve Analysis (DCA) plot evaluates the clinical usefulness of a predictive model. The X-axis shows the threshold probability for taking action, while the Y-axis represents the standardized net benefit. The blue curve represents the model’s net benefit across different thresholds, compared to the red lines indicating net benefits if everyone (solid red) or no one (dashed red) were treated. The green shaded region highlights thresholds where the model provides a positive net benefit, while the red region shows where the benefit decreases. (B) This graph shows the relationship between diagnostic metrics (FNR, FPR, NPV, PPV) and risk thresholds for hospital admission. As the threshold increases, the False Negative Rate (FNR) incasese and False Positive Rate (FPR) decrease, while Positive Predictive Value (PPV) increases and Negative Predictive Value (NPV) decreases. The green and red shaded areas highlight threshold ranges where these metrics are optimized for clinical decision-making