There is an easy fix but nobody would stand for it: do full longitudinal surveillance of all tests. That is, the gent with the “narrowed” arteries could be followed to see if it becomes a problem (a lot of the metrics are, to simplify, fairly arbitrary and based on a small observational n). Perhaps he does have a stroke, perhaps not.

Of course no one would stand for such an intrusive system. But over time it would improve diagnosis.