Serving the Quantitative Finance Community

 
User avatar
katastrofa
Posts: 7430
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Models for Covid-19

August 7th, 2020, 6:52 pm

Pressure or temptation (with bribes?). I struggle to understand the message - unless it comes from people with no moral spine.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

August 17th, 2020, 3:03 pm

Tory MP husband of Test and Trace chief Dido Harding linked to anti-NHS group

https://www.mirror.co.uk/news/politics/ ... f-22128892
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

August 22nd, 2020, 4:58 pm

Harvey Stein on COVID-19 data

https://hjstein.blogspot.com/2020/05/co ... in_33.html

I find it surprising that epidemiologists who make a career out of analyzing epidemics and pandemics would record the data in such a fashion. But, on the other hand, I suppose such work tends to be on a longer time scale and it’s only in the current pandemic that we needed accurate, up to date infection and death counts.


I’d hope that someone would take it upon themselves to collect and aggregate the data on an incident date basis. This would be a huge undertaking, but the longer this pandemic persists, the more important this becomes.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 7th, 2020, 6:48 pm

'We are heading into another epidemic' warns Prof McConkey

https://www.rte.ie/news/health/2020/090 ... l-figures/
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 9th, 2020, 10:44 am

What "Second wave" SEIR models?
https://arxiv.org/pdf/2006.05081.pdf

As an armchair epidemiologist (aren't we all?) I always asked about the E in SIR.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 9th, 2020, 10:49 am

N=Population
S=Susceptible
E=Exposed
I= Infectious
R=Removed (or Recovered)
D=Death
C=Control measure(s) applied (?) just a guess here, but the NPI interventions and combinations of interventions they applied

Some level of confirmation from a similar paper with some clearly explained variables.
https://www.thelancet.com/journals/lang ... 7/fulltext

Paper summary, pdf and shared R code via github (but this one has simulations only - not real world data)

Also see:

https://www.imperial.ac.uk/news/196234/ ... ly-impact/
Select Researchers/coauthors listed and quoted there. 
yes, E looks important.

Anecdotal evidence suggests they use SEIR model with time-varying parameters in China.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 9th, 2020, 10:56 am

Covid-19: AI model confirms at 98.5%: there will be no second wave this summer or this winter According to Devoteam’s data scientists

https://www.devoteam.com/newsroom/covid ... cientists/

This study is based on hybrid regression approaches coupled with reinforcement learning Bandit models and the Bayesian SEIR model [1]. It takes into account primary and primary-secondary symptoms of COVID-19 and data reported by the French emergency departments for visits or calls. 

some kind of charged-up State Machine??
 
 
User avatar
katastrofa
Posts: 7430
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Models for Covid-19

September 11th, 2020, 6:35 pm

Several months ago I got obsessed with a theory that the severe covid is the immune enhanced reaction to corona. I remember reading research on the cytokine storm claiming that it can be triggered by bradykinins.
I've never been enthusiastic about poorly researched vaccines against poorly researched diseases, but this and the discovery of new proteins (I love the tone of this report - "yay! it's potentially way nastier and more robust that we thought!") make me even more suspicious.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 12th, 2020, 7:02 pm

Haven't heard much about the Cube Model lately.

rump’s advisers released a ‘beyond stupid’ mathematical model of coronavirus deaths created in Excel by a controversial economist

https://www.businessinsider.nl/trumps-c ... =true&r=US

“Hassett remains … a pillar of the modern conservative establishment, and Trump called on him to second-guess experts in epidemiology, a field in which he has no background,” Krugman wrote.

Earlier this week, The Washington Post reported that the White House is relying in part on a so-called “cubic model” devised by Kevin Hassett, a top economic adviser, that shows COVID-19 deaths plummeting to zero by mid-May.

This model, which the White House Council of Economic Advisers (CEA) released on Tuesday, contradicts all other data and modeling, including that the White House is reportedly looking at. The University of Washington’s Institute for Health Metrics and Evaluation predicts 135,000 deaths by August 1. Nearly 73,000 people have already died from the disease in the US, with about 2,000 dying every day over the last week.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 22nd, 2020, 11:17 am



There is more to life than ODEs but my question is why these reports do not discuss these problems as dynamical systems ("ODEs++") so that other questions can be answered beyond numerical ODEs. I wonder  why no mention is made of the Lyapunov, Poincare-Bendixson theory etc. What happens to the ODEs when [$]t \rightarrow \infty[$] (sounds like a logical question). A Sapir-Whorfe problem ("give hammer then all is a nail")
Second wave?
I did warn you; conventional wisdom was a combination of desperately seeking sigma curve and herd immunity.
 
User avatar
katastrofa
Posts: 7430
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Models for Covid-19

September 22nd, 2020, 10:29 pm

There is more to life than ODEs but my question is why these reports do not discuss these problems as dynamical systems ("ODEs++") so that other questions can be answered beyond numerical ODEs. I wonder  why no mention is made of the Lyapunov, Poincare-Bendixson theory etc. What happens to the ODEs when [$]t \rightarrow \infty[$] (sounds like a logical question). A Sapir-Whorfe problem ("give hammer then all is a nail")
They do. Many epidemiologists use network models of contagion / directed percolations. The critical behaviour is at the threshold between the active and the absorbing regime (cf R0).
Funny, in the other forum I saw some Jan Dash posting papers on Regge field theory (from high energy physics, any CERN people here to give more insight?) as "generalisation" of a Brownian motion. RFT, Brownian motions and the network model of epidemic/directed percolation mentioned above are in the same universality class - every model with an underlying stochastic Markov process qualifies to this class (every reaction-diffusion system: contagion, reacting chemical substances, 2nd order phase transitions like ferromagnetic - note percolation threshold vs critical, and so on. Scientists are copycats...

Anyway, let's assume a naive model the contagion as an isotropic percolation, i.e. Bethe lattice:
Image

A classic result is that large clusters and long-range connectivity arise when the probability of "jumping" to the next node is Pc = 1/(z-1) (*), where z is the coordination number. In our model the latter corresponds to the number of regular contacts made by an infectious (it's a static model - contacts don't change).
Assume that for Covid the probability of "jumping", i.e. infecting the contact, is 15% per each meeting. This probability is due to the Poisson process with intensity given by the equation 0.15 = 1-exp(-lambda*1), which yields lambda = -ln(0.85).
Say the disease period is 21 days and you meet all your contacts once a day. Then the probability that you infect each contact accumulates to P(21 days) = 1-exp(ln(0.85)*21) = 1 - 0.85^21. For this probability to be the percolation/epidemic outbreak threshold, it takes z susceptible contacts given by Eq (*), which is 2.0340679919043606448571830240222 persons.
Don't tell Boris!
Last edited by katastrofa on September 23rd, 2020, 8:44 am, edited 1 time in total.
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 23rd, 2020, 8:36 am

What's the difference between percolation and diffusion? How do you model individuals distributed in space?
Of course, this model has little to do with (spatial) SIR?

I feel the above is a case of argumentum ad verecundiam.
 
User avatar
katastrofa
Posts: 7430
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Models for Covid-19

September 23rd, 2020, 11:16 am

One can look at it in different ways from the perspective of mathematics, but I think the answer for you is that in a classic diffusion model you can go in either direction with an equal probability. In a percolation model you can access some sites with a certain probability. Vide anomalous diffusion.

@"I feel the above is a case of argumentum ad verecundiam."

It probably means that you need to educate yourself a bit on the subject(s) if you want to discuss them ;-)
 
User avatar
Cuchulainn
Topic Author
Posts: 20204
Joined: July 16th, 2004, 7:38 am
Location: 20, 000

Re: Models for Covid-19

September 24th, 2020, 12:08 pm

Swedish expert backtracks on herd immunity for Ireland

https://thestreetjournal.org/2020/09/sw ... r-ireland/

He said schools must remain open and he pointed out that there was no difference between infections among schoolchildren in Sweden, where schools were kept open, and in neighbouring Finland which closed its schools.

“What you just told me, I’m unaware of this. I’m objecting to you saying I’m telling Ireland what to do. I’m not. 

Speaking later, Dr Giesecke denied that he was prescribing the Swedish model for Ireland after it was put to him that differences between Ireland and Sweden’s population densities, classroom sizes, and sick pay entitlements would likely mean different outcomes in the two countries.

//
Aka, backtracking, no true Scotsman fallacy. Awful. 

"The Ireland example is no true Finland example."
Last edited by Cuchulainn on September 24th, 2020, 1:01 pm, edited 6 times in total.