It is not appropriate to seek a cutoff for a test unless every patient is
known to have exactly the same utility function, which is extremely rare.
There is nothing wrong with developing probability models and stopping at
that point, deferring the decision to the possessor of the utility function.
Dear all,
I would like to calculate the optimal cut off (threshold) of a test using
the Epi package. Here I am presenting some data based on the output of two
tests. I am interested in identifying the optimal cut off an its 95% CI.
Running the ROC() function with the Epi package I obtain a nice
2 matches
Mail list logo
