Skip to main content
Fig. 1 | BMC Research Notes

Fig. 1

From: The clinical utility curve: a proposal to improve the translation of information provided by prediction models to clinicians

Fig. 1

ROC curves (a) and CU curves (b) for the prediction of malignancy (combined malignancy and borderline histology) in 37 women with adnexal masses. Solid line represents model 1 and dashed line represents model 2. NNCOC number of surgical biopsies needed to capture one case of malignancy. C statistics were 0.839 (95 % CI 0.697–0.947) for model 1 and 0.877 (0.753–0.968) for model 2, P = 0.003. Sensitivities of 80, 87, 93 and 100 % were achieved at absolute risks of 38, 36, 32 and 3.7 %, respectively, for model 1, and 64, 30, 12 and 6 %, respectively for model 2. Model 1 identified all 15 malignancies in 35 patients, with an NNCOC of 2.3, whereas model 2 identified all 15 malignancies in 30 patients, with an NNCOC of 2

Back to article page