Serving the Quantitative Finance Community

  • 1
  • 2
  • 3
  • 4
  • 5
  • 34
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Models for Covid-19

March 17th, 2020, 8:46 pm

Today on BBC they actually showed the (udocumented) ODE model (an ODE system with 7 unknowns). Here is a snippet; Can it model Corona? sound too good to be true. And almost linear. My guess is that it cannot model a lot? It looks awful linear, almost. 1st year electrical engineering syllabus??

Image

Paul? Alan? Farid? Kats
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 17th, 2020, 9:03 pm

There’s a little bit of nonlinearity in there.

What are S, I and E?

I bet it gives a sigmoid all shape for things. Every model does. But which is the right one?!

Seven unknowns? Do you mean variables or parameters?
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

March 17th, 2020, 9:35 pm

Seven ode unknowns like (Ok, it's a bit nonlinear but still)

dv1/dt = ...
dv2/dt = .
..
dv7/dt = 

They will probably show it again on 10 o clock BBC news. They actually mentioned what it could not model.

It seems to be from Imperial. it would be interesting to see the full model, 
Last edited by Cuchulainn on March 17th, 2020, 9:39 pm, edited 1 time in total.
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 17th, 2020, 9:38 pm

Where’s the paper?

Those three look self contained. Unless N (which I assume to be population) is from another ode because of lots of deaths!
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

March 17th, 2020, 9:40 pm

Where’s the paper?

Those three look self contained. Unless N (which I assume to be population) is from another ode because of lots of deaths!
dN/dt was the first ODE on the BBC screen. I'll try to find it. AFAIK 'N' is usually the population.
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 17th, 2020, 9:43 pm

That’s not a good sign!
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

March 18th, 2020, 9:03 am

I managed to quickly write down  the dependent variables for what its worth: N, S, E, I, R, D and C (i have no idea what they mean). Anyways, the researchers at Imperial (Neil Ferguson et al) have stark mortality rate predictions based on that ODE model.
I have been unable to locate the complete article.
 
User avatar
trackstar
Posts: 3420
Joined: January 1st, 1970, 12:00 am

Re: Models for Covid-19

March 18th, 2020, 11:45 am

Where’s the paper?

Those three look self contained. Unless N (which I assume to be population) is from another ode because of lots of deaths!
Here is the paper - maybe contact the researchers to see if they will provide the full details on the model?

Imperial College report:

https://www.imperial.ac.uk/media/imperi ... 3-2020.pdf
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

March 18th, 2020, 12:25 pm

What they should do is make model, data and assumptions 100% explicit so that others can reproduce the results in a COMPUTER.
 
User avatar
trackstar
Posts: 3420
Joined: January 1st, 1970, 12:00 am

Re: Models for Covid-19

March 18th, 2020, 12:50 pm

What they should do is make model, data and assumptions 100% explicit so that others can reproduce the results in a COMPUTER.
Link to appropriate center and contacts:

MRC Centre for Global Infectious Disease Analysis - COVID-19 reports

 On behalf of the Imperial College COVID-19 Response Team

WHO Collaborating Centre for Infectious Disease Modelling
MRC Centre for Global Infectious Disease Analysis
Abdul Latif Jameel Institute for Disease and Emergency Analytics
Imperial College London


They may be overwhelmed with responses and requests, but one of you can try.
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 18th, 2020, 3:13 pm

Where’s the paper?

Those three look self contained. Unless N (which I assume to be population) is from another ode because of lots of deaths!
Here is the paper - maybe contact the researchers to see if they will provide the full details on the model?

Imperial College report:

https://www.imperial.ac.uk/media/imperi ... 3-2020.pdf
i haven’t been able to find the full model, only the chat about results.

And Neil Ferguson is now self isolating with symptoms.
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 18th, 2020, 3:21 pm

I managed to quickly write down  the dependent variables for what its worth: N, S, E, I, R, D and C (i have no idea what they mean). Anyways, the researchers at Imperial (Neil Ferguson et al) have stark mortality rate predictions based on that ODE model.
I have been unable to locate the complete article.
N population?
S susceptible?
I infected or infectious?
R removed or recovered?
D deaths?

They need something about deaths due to damage to the economy in there. That’s looking very bad.

The model is probably useless. It’s going the same way as economics models. We are still in back of the envelope time, ie v simple but robust. Forget large numbers of coupled odes.
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 18th, 2020, 3:48 pm

Find the model and the parameter values. Then we’ll tweak it to allow for deliberate staggered infection of the young. See what happens to the results. Simples. Cuch, you’re on the numerics!
 
User avatar
Cuchulainn
Topic Author
Posts: 20253
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

March 18th, 2020, 4:24 pm

I'm also not sure an ODE model is good (indeed,didn't you mentioned that economics analogy  once a while back 700 parameters?) , even with the parameters. Of course, if it is not public domain ....
 
User avatar
Paul
Posts: 6604
Joined: July 20th, 2001, 3:28 pm

Re: Models for Covid-19

March 18th, 2020, 4:47 pm

Exactly.

It will be out there somewhere.

All we need to do is change the initial conditions.