Figure 4 | Scientific Reports

Figure 4

From: Combination of clinical information and radiomics models for the differentiation of acute simple appendicitis and non simple appendicitis on CT images

Figure 4

Cross-validation in the training dataset and ROC curves in the test dataset. (A) The process involved fivefold cross-validation in the training set for the three models, using the same data subdivision. The AUCs of the CT model were as follows: 0.675, 0.698, 0.663, 0.664, and 0.697, with an average AUC of 0.679. The AUCs of the radiomic model were 0.664, 0.852, 0.639, 0.712, and 0.750, with an average AUC of 0.723. The AUCs of the combined model were 0.883, 0.885, 0.891, 0.881, and 0.889, with an average AUC of 0.886. (B) ROC curves of the CT model, radiomics model, and combined model in the test dataset. There was no significant difference in the AUCs between the combined model and the radiomics model according to the DeLong test (0.817 vs. 0.804, P = 0.808). The AUC of the combined model was significantly higher than that of the CT model (0.817 vs. 0.669, P = 0.041). The AUC of the radiomics model was also higher than that of the CT model but did not reach a level of statistical significance (0.804 vs. 0.669, P = 0.053).

Back to article page