> They could predict with a probability of around 90 percent that individual patients would later become seriously ill or die from the Covid-19 infection. The p

> They could predict with a probability of around 90 percent that individual patients would later become seriously ill or die from the Covid-19 infection. The prediction of a less severe course was somewhat less precise. That is, in individual cases, a predicted severe course did not become evident. I wonder if this would correlate with date of infection vs date of testing. As in, a high expression of hFwe-Lose in the first N days after infection is a reliable indicator, but if the testing is done at a later stage, the baseline hFwe-Lose expression is higher in all patients and therefore no longer an adequate predictor. I wonder if there is a second biomarker that can provide an indication how long the infection has been progressing, and if they could correlate those two biomarkers? The linked paper (https://news.ycombinator.com/item?id=29263350 https://news.ycombinator.com/item?id=29263350) doesn't seem to mention this at all, perhaps because it's impossible to determine. Looking at the sensitivity-vs-selectivity charts, it may simply be that this biomarker by itself casts too wide a net.