Radiomics signature: A potential biomarker for β-arrestin1 phosphorylation prediction in hepatocellular carcinoma

Figure 4 Decision curve analysis for each model. A: Decision curve analysis in the training cohort; B: Decision curve analysis in the validation cohort. The y-axis measures the net benefit, and the x-axis is the threshold probability. The gray line represents the hypothesis that all patients are β-arrestin1 phosphorylation-positive. The black line represents the hypothesis that all patients are β-arrestin1 phosphorylation-negative. Among the three models, the clinico-radiological-radiomics (CRR) model provided the highest net benefit compared with the radiomics and clinico-radiological (CR) models.