|| "The choice of values for test
sensitivity, specificity and predictive value and ultimately, the definition of the
“positivity criterion” depend on a decision whether it is worse to label a patient
diseased when he is not (false positive) or to label him disease-free when he is not
(false negative). This tradeoff should take into account the needs of the patient as well
as the judgement of the physician" (Knapp and Miller, 1992).Scenario:
1. When the consequences
of missing a case are potentially grave.
2. When a false positive
diagnosis may lead to a risky treatment.
characteristic (ROC) curves are used to compare the performance of two competing
diagnostic tests and to determine the appropriate cutoff values of those tests. The
technique is used when you have a continuous criterion (predictor or independent) variable
which will be used to make a yes or no decision. ROC curves are most often employed in the
medical fields. It may be used with any classification procedure. For example, in may be
used with two-group discriminant analysis to help determine the appropriate cutoff value
of the discriminant score for classifying a individuals (Knapp and Miller, 1992).