Value of 18F-FDG-PET/CT radiomics combined with clinical variables in the differential diagnosis of malignant and benign vertebral compression fractures

EJNMMI Research

Table 2 Performance of the prediction models

Prediction model	AUC	95%CI	Accuracy	Sensitivity	Specificity	PPV	NPV	Group
Clinical model	0.884	0.8153–0.9518	0.850	0.830	0.872	0.880	0.820	Train
Radiomics model	0.986	0.9714–1.0000	0.940	0.887	1.000	1.000	0.887	Train
Clinical–radiomics model	0.987	0.9716–1.0000	0.950	0.906	1.000	1.000	0.904	Train
Clinical model	0.858	0.7437–0.9729	0.841	0.875	0.800	0.840	0.842	Test
Radiomics model	0.962	0.9137–1.0000	0.932	0.917	0.950	0.957	0.905	Test
Clinical–radiomics model	0.948	0.8787–1.0000	0.932	0.958	0.900	0.920	0.947	Test

AUC area under curve; CI confidence interval; PPV positive predictive value; NPV negative predictive value