Biomedical Engineering Reference
In-Depth Information
Another way to evaluate the sensitivity and specificity of a statistical test is to determine its receiver
operating characteristic (ROC) curve, as in
Figure 6-21
. The ROC curve is a plot of a test's sensitivity
versus 1 - specificity, or true-positive rate versus false-positive rate. The higher the curve of a test,
the greater its discriminative ability. Every point along an ROC curve corresponds to test sensitivity
and specificity at a given threshold of abnormal. All else being equal, a test with the greatest
discriminative ability (Test A) is superior to a test with lower discriminative ability (Test B).
Figure 6-21. Receiver Operating Characteristic (ROC) Curves For the two
tests shown here, Test A provides superior discrimination over Test B.
