[BostonCard]

I've seen at least 3 or 4 models that predict fairly high R0 values (like >4), including the one from Imperial College and one that was in PNAS this week, I think.  Those R values lead to the prediction of a high number of infected people, most of which would have to be asymptomatic or have mild symptoms to match up with observed numbers of hospitalizations and deaths.  

I've been dismissive of those models as wishful thinking, but we won't really know until the antibody surveillance comes in.

Do these antibody test numbers account for false positives and negatives?  If you know the sensitivity and specificity of a test, it should be trivial to correct for both, right?

You can calibrate the specificity by looking at pre-2020 banked plasma.  The prevalence of antibodies should be zero back then.  Sensitivity can be established by testing known Covid-19 (with positive PCR tests) after convalescence.  And yes, then it should be possible to adjust.

BC

[BostonCard]

I've seen at least 3 or 4 models that predict fairly high R0 values (like >4), including the one from Imperial College and one that was in PNAS this week, I think.  Those R values lead to the prediction of a high number of infected people, most of which would have to be asymptomatic or have mild symptoms to match up with observed numbers of hospitalizations and deaths.  

I've been dismissive of those models as wishful thinking, but we won't really know until the antibody surveillance comes in.

Do these antibody test numbers account for false positives and negatives?  If you know the sensitivity and specificity of a test, it should be trivial to correct for both, right?

You can calibrate the specificity by looking at pre-2020 banked plasma.  The prevalence of antibodies should be zero back then.  Sensitivity can be established by testing known Covid-19 (with positive PCR tests) after convalescence.  And yes, then it should be possible to adjust.

BC