FIGURE 2-3 ROC curves representing accuracy index values (A) of 0.5, 0.7, 0.8, and 0.9.

tests. Reading across from the axis at the right of Figure 2-3 to the one at the left, one can see that for any fixed rate of correct identification of positive cases (sensitivity), the more accurate the test, the smaller the proportion of truly negative cases incorrectly judged positive, read from the axis at the bottom. Similarly, for any false positive rate, shown on the axis at the bottom of the figure, the more accurate the test, the greater the proportion of positive cases that are accurately identified.8

Decision Thresholds

Figure 2-2 shows three points corresponding to different thresholds on a curve with A = 0.8. The point B is the balanced threshold, meaning

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement