You must log in or register to comment.

celebratedrecluse wrote (edited )

This is a MosaiQ serological screening machine. So it's a lab test, not a portable or at-home one. I'm not sure how long it takes to process data, but it can operate in the cloud so hopefully places with internet access will be able to accept samples so this can be used more widely. Right now they have 12 and another 20 will be made by the end of the year, but that obviously is a paltry number. With 35 minutes per sample, I estimate each machine can process about 45 tests per day, unless they can be run in parallel. (I don't think so, but I'm not sure.) This amount to just a few thousand tests per day, which is hardly sufficient to test the majority of the country, so it will unfortunately be limited just to healthcare sector applications rather than the epidemiological survey that was needed as of months ago. Ergo, if you get sick and end up hospitalized, they might test you once you start to recover to see if you are immune and a candidate to help the vaccine development process.

99.8% specificity means that false positives may occur up to 2 in every thousand. When you are testing the majority of a population of millions, this can be a significant margin of error (tens, even hundreds of thousands of false positives). This means that some people are going to be told they are immune to the virus when they actually don't have the antibodies which signal immunity. Something to keep in mind.

On the plus side, if the sensitivity is actually 100% (probably just >99.99% at most, tbh) this means that almost all people with immunity will be identified, which is good. Only a few thousand people in the UK with immunity will be told they do not have immunity when they actually have immunity, if the whole UK gets this test.


masque wrote (edited )

Possibly a silly question: in order to measure the specificity & sensitivity of the test, how do they find the ground truth about which patients do or don't have the antibodies prior to the development of the test for antibodies? I guess you could use a slower but more reliable test, but is there then some sort of chicken & egg problem?


celebratedrecluse wrote

an interesting epistemology question, which I am not qualified to answer. Perhaps someone else can chime in with more information and knowledge of this field, I would be very interested to learn about it