How do you assemble false positive false negative statistics for a new diagnostic in the presence of exponential growth ...?

Is this actually possible to do — my instincts say that any exponentially transmitting contagion will make it impossible to assess such stats with any confidence for quite some time ...