The prototypical example of the problem of cost-sensitive classification is medical diagnosis,
where a doctor would like to balance the costs of various possible medical tests with the
expected benefits of the tests for the patient. There are several aspects to this problem: When
does the benefit of a test, in terms of more accurate diagnosis, justify the cost of the test?
When is it time to stop testing and make a commitment to a particular diagnosis? How much
time should be spent pondering these issues? Does an extensive examination of the various
possible sequences of tests yield a significant improvement over a simpler, heuristic choice
of tests? These are some of the questions investigated here.