Figure 3.

Development and internal validation of the hospital survival and transplant-free survival nomograms. (A) The hospital survival nomogram. The nomogram was based on four prognostic factors and could be used to predict the probability of hospital survival at 15, 30, and 45 days by adding up the points identified on the scales of these four parameters. (B) The time-dependent ROC curves and AUC at 15, 30, and 45 days are shown. (C) The Brier score calibration curves for the hospital survival nomogram at 15, 30, and 45 days. (D) The transplant-free survival nomogram. The nomogram was based on five prognostic factors and could be used to predict the probability of transplant-free survival at 6, 12, and 18 months. (E) The time-dependent ROC curves and AUC at 6, 12, and 18 months are shown. (F) The Brier score calibration curves for the transplant-free survival nomogram at 6, 12, and 18 months. Internal validation of the two nomograms was performed using the bootstrap sampling method. The time-dependent ROC curves were measured by bootstrapping with 1000 repetitions. A calibration curve developed using the bootstrap method with 1000 repetitions was used to estimate the probability at different times. The X-axis represents the predicted probability calculated from the nomogram, and the Y-axis represents the actual probability. Abbreviations: PTA: prothrombin activity; CKD: chronic kidney disease; AKI: acute kidney injury; sCr: serum creatinine; MELD–Na: model for end-stage liver disease–sodium; CI: confidence interval; AUC: area under the receiver operating characteristic curve; ROC: receiver operating characteristic.