Skip to main content
. 2022 Apr 13;12:6193. doi: 10.1038/s41598-022-10136-9

Figure 4.

Figure 4

Cross-validated performances of the developed models. The best-performing classifier for the imaging features was a support vector classifier using 2 features with a linear kernel, achieving a 0.88 AUC. The mortality score model obtains a 0.62 AUC with a logistic regression model. There are 152 positive samples and 126 negative samples in the validation set.