Vinay Prasad

And you say, in the best case scenario, do you have still problems?

And you can see you're going to have a significant number of people who are in the false positive and the false negative group.

And in fact, they cancel each other out.

Well, it doesn't matter if you're using it for random sampling.

That's what I was trying to get to.

If you work in a research study and you want to work out how many people you think have got the disease, you can factor in that effect.

But if you want to give people advice and say, oh, by the way, you've got antibodies and you're definitely protected, you have a problem with that result because actually 30% of the time you're not going to be protected.

And I'm picking my words carefully here because I'm sort of thieving at the edges here.

I thought they were numbered on the basis it was a sort of blinding issue, which makes complete sense.

But the idea with signing up to studies and producing evidence on pieces of kit that you can't actually name the kit and what it is, is just not on.

And what it shows is a complete loophole in the device industry, in the approval of devices, and the complete mess that is diagnostics, but also medical devices across the board.

Yes, there is no central database for devices and technologies.

There's supposed to be one called Udamed.

And believe it or not, it's supposed to be fully functioning next month in May.

and about and somebody will correct me here it's supposed to be fully functioning on the 21st 22nd of may but in the new regulations it says if it's not fully functioning by that date he can revert back to the old regulations and push it back therefore there is no single database of devices that's going to appear anytime soon and

If this wasn't so serious, it would be a joke.

But it is really serious.

And I think we need a rewriting of the whole script here.

Because you cannot have a situation where you try all technologies, drugs, and you keep what it is hidden.

Because what happens if that's shown to be harmful?