Criterion-free diagnostics
Researchers are often (correctly) discouraged from reporting measures of
diagnostic performance derived from a single, arbitrary criterion, and
should rather describe the test’s performance across all possible
criteria. For these purposes, researchers plot the receiver operating
characteristic (ROC) curve, from which the area under the curve (AUC) is
derived and provides a non-parametric, criterion-free estimate of the
test’s sensitivity. When deriving the AUC for BSITER scores, we suggest
using a large number of iterations (e.g., 10,000). Or, if one has
already conducted the rBS test, the average BSITER score from this test
can be used.