Multiphase CT-based prediction of Child-Pugh classification: a machine learning approach

European Radiology Experimental

Table 4. Accuracy scores for all predictive models

Predictive model	LR	RF	CNN	ERs
Prediction of the Child-Pugh class
Spearman ρ	0.35	0.32	0.51	0.60
Accuracy (%)	47	47	53	57
Classification Child-Pugh class A versus ≥ B
Accuracy (%)	71	70	78	71
Sensitivity (%)	85	81	81	82
Specificity (%)	38	43	70	66
AUC	0.71	0.67	0.80	0.76

AUC Area under the curve at receiver operating characteristic analysis, LR Linear regression, RF Random forest, CNN Convolution neuronal network, ERs Experienced radiologists