that drops the total number of false positives from 49500 to 39600, which is still 80% of the positive results being incorrect

our learned doctor friend is correct, we haven't made use of tests available to date because Bayesian stats magnify any inaccuracy, but the tests we're talking about now are accurate/specific to the point where that is an acceptable/useful level - particularly if they are generating an extremely small number of false negatives, which are obviously more important