Skip to main content
Fig. 2 | Radiation Oncology

Fig. 2

From: Machine learning-based normal tissue complication probability model for predicting albumin-bilirubin (ALBI) grade increase in hepatocellular carcinoma patients

Fig. 2

Behaviors of the best gradient-boosted tree model. The impact of each input feature on the prediction was estimated empirically by altering the feature value and recording the corresponding change in the model’s output. Red trend lines show the average relative change in prediction. Blue shaded areas indicate the plus/minus one standard deviation range. (a) Impact of the ALBI score until next grade on the prediction. (b) Impact of the baseline ALBI score on the prediction. (c) Impact of the baseline AST on the prediction. (d) Impact of the baseline total bilirubin on the prediction. (e) Impact of normal liver volume on the prediction. (f) Impact of the gEUD calculated at a = 1.0, or mean liver dose, on the prediction

Back to article page