Skip to main content
. 2023 Apr 10;9(4):e15283. doi: 10.1016/j.heliyon.2023.e15283

Fig. 2.

Fig. 2

Discrimination ability of the prediction models for death and critical outcomes in patients with COVID-19 in derivation cohort (Panel A and B) and in validation cohort (Panel C and D). Panel A represents the ROC areas for death in the derivation cohort with a shrunken AUC of 0.85. Panel B represents the ROC areas for critical outcomes in the derivation cohort with a shrunken AUC of 0.79. Panel C represents the ROC areas for death in the validation cohort with an AUC of 0.85 (95%CI, 0.80–0.90). Panel D represents the ROC areas for critical outcomes in the validation cohort with an AUC of 0.78 (95%CI, 0.73–0.83). The diagonals represent that the models have no discrimination with an AUC of 0.50. ■ represents the optimal cutoff for the predicted probability where the sum of sensitivity and specificity is the maximum.