Let's consider a case of a test for presence/absence of a viral infection. When we have certainty as to whether a specific individual is carrying the virus, it is easy to estimate the False Positive Rate (FPR) and the False Negative rate (FNR) of a new diagnostic test. However, I suspect that for many real case scenarios, we cannot have certainty. In such case, I can think of two general (non-exclusive) solutions to estimate the FPR and the FNR of a test.
- Cross validation with same test
- Cross validation with other test (such other test may include a boundary threshold based on symptoms and signs)
Are there other techniques to estimate FPR and FNR of a diagnostic test? In general, could you point me to a source that discuss such techniques?