Figure 3

Receiver Operator Characteristic curves. A. Receiver Operator Characteristic (ROC) curves were generated by plotting the sensitivity (i.e. the proportion of times the test correctly predicts a patient as metastatic) against 1-specificity (i.e. 1-the proportion of times the test correctly predicts a patient as non-metastatic) as a measure of the predictive value for each test. Shields Index calculated by total LD and by hot spot, LD, Breslow thickness, Breslow thickness for the 4500 patients included in the Frenchay Hospital Melanoma Registry and AJCC staging. B. The area under the ROC curve (AUC) represents numerically the accuracy of the test. A prognostic indicator with 100% sensitivity and 100% specificity has an AUC of 1.0. The most accurate prognostic indicator is the Shields Index, followed by epi-tumoural LD, Breslow thickness and finally AJCC staging