Thread by @fascinatorfun

Coronavirus exposes the problems and pitfalls of modelling | Science | The Guardian

This takes some of the unhelpful heat out of the Oxford Uni study that seeks to set potential parameters for a model, then test it to see where the truth lies.
theguardian.com/science/2020/m…

No one knows what fraction of the public is at risk of serious illness
The study demonstrates how wildly different scenarios can produce the same tragic pattern of deaths & asserts the urgent need for serological testing for antibodies against the virus, to discover which is true

The way hypothesis - a state of affairs to be tested - has been seized on as statements of truth within the press has been unfortunate

It makes the point that serological testing is critical.

Of course it cannot establish the scale of the pandemic. Merely point to a direction .

Critically always it has to be remembered that a model is just that.

They are based on a set of assumptions & those assumptions have to be constantly reviewed in the light of changing information.

So early Feb the WHO said asymptomatic CV19 likely time be the 1-3%
No longer

Now several studies and emerging data from wide testing (China, S Korea, Japan, Royal Princess, Iceland) suggest that asymptomatic or symptoms so mild they went unnoticed forms a much bigger subset than previously thought 25-33%. And is transmittable

That has huge implications for any model.

In addition the assumption that the RO is 2.3 (i.e. the number of people likely to be infected by one transmitter) is also in doubt.

We are in a large world of “don’t know” and testing out the parameters of any hypothesis by serology testing is a start to eating into the “don’t knows”.

It may well blow a hole in the Government modeller assumptions. It may bear them out.

Quite likely it may land in between