Carl Hennigan

But in somebody who's 95, who their sats are 94 all the time, you might go, well, actually, that's just what they are.

And that's where they are.

And actually, you wouldn't act on that person.

And so that's the problem with a one size fits all prediction.

Well, let's take the biggest disease in the world, cardiovascular disease, where we've done the most evidence and the most prediction rules and looked at it in so many ways.

They're very interesting, but there's no evidence to show that if you use prediction rules, A, that they actually predict the people who actually have the disease, and B, using them improves overall mortality.

And I think you have to remember these sorts of things when you're trying to go, actually, if we just say, take this model and ship all these people in,

Is that what you're actually going to do?

Well, then you work out, why do I have doctors?

Why do I have any in the process?

They're not the decision-making tool.

They are the communicating tool that allow us.

So when I use this new score, it's I've used it because I've already made my decision about this patient needs to go in hospital.

I'm using my clinical skills, other information, where we are.

And you're right.

And what you want to know is, is this person likely to benefit from the next piece of management?

Now,

Where's the evidence for that?

There is none, is there?

So it's not just the prediction rule.