Fig. 4From: Interpretable recurrent neural network models for dynamic prediction of the extubation failure risk in patients with invasive mechanical ventilation in the intensive care unitCalibration curves for model validation on the test set. For each model, the calibration curve plotted means of decile-binned predicted probabilities versus corresponding means of actual probabilities in the patients in each bin. As shown, each blue point of a calibration curve represented a bin and the size of the gray circle around represented the sample size of this bin. The dotted line was the identity line of y = x representing perfect calibration. Abbreviations: GRU gated recurrent unit, LSTM long short-term memory, LR logistic regression, MLP multi-layer perceptron, RF random forest, SVM support vector machine, XGB extreme gradient boostingBack to article page