Fig. 1From: Machine learning-based normal tissue complication probability model for predicting albumin-bilirubin (ALBI) grade increase in hepatocellular carcinoma patientsImportance of each input feature for the best model. (a) Logistic regression model. Scatter plots of the coefficient values from the five logistic regression models trained using different partitions of the 5-fold cross-validation. Features with higher absolute magnitudes are considered more important to the prediction. Red bars indicate the average values. (b) Random forest model. Scatter plots of the Gini importance values from the five random forest models trained using different partitions of the 5-fold cross-validation. Features with higher Gini importance score are considered more important to the prediction. Red bars indicate the average values. (c) Gradient-boosted tree model. Scatter plots of the Gini importance values from the five gradient-boosted tree models trained using different partitions of the 5-fold cross-validation. Red bars indicate the average valuesBack to article page