Figure 3 | Scientific Reports

Figure 3

From: Personalized prediction of delayed graft function for recipients of deceased donor kidney transplants with machine learning

Figure 3

Calibration plots of the uncalibrated (red) versus recalibrated (blue) models. The plots show the observed prevalence of delayed graft function (DGF) versus predicted probability of DGF per decile. The Hosmer–Lemeshow test was performed to test for calibration errors. Only the baseline model had good calibration without recalibration (p > 0.05). When recalibrated with Platt scaling, all of the models showed improvement. BL baseline, LR logistic regression, EN elastic net, RF random forest, XGB extreme gradient boosting, ANN artificial neural network.

Back to article page