Implications of Heterogeneous SIR Models for Analyses of COVID-19

This paper provides a quick survey of results on the classic SIR model and variants allowing for heterogeneity in contact rates. It notes that calibrating the classic model to data generated by a heterogeneous model can lead to forecasts that are biased in several ways and to understatement of the forecast uncertainty. Among the biases are that we may underestimate how quickly herd immunity might be reached, underestimate differences across regions, and have biased estimates of the impact of endogenous and policy-driven social distancing.

That is the abstract of a new paper by Glenn Ellison, recommended.

Comments

We're going to need a natural experiment/

Respond

Add Comment

Even more recommended is the free download provided by Springer of Epidemiological Research: Terms and Concepts at https://link.springer.com/book/10.1007%2F978-94-007-1171-6

Respond

Add Comment

In other words, the lockdown was a great success. And people are mad as Hell about it! Go figure. Yesterday VP Pence advised governors to "explain to your citizens" that it's only increased testing that has caused the spike in new cases and hospitalizations. The lesson taught Citizens of Oz by the Wizard of Oz is that what people believe is what is. Thus, the cowardly lion who believes he is brave is; the idiot scarecrow who believes he is smart is; and the heartless tin man who believes he is a sensitive soul is. Our world is little more than an illusion, illusion manufactured from thin air; or in Pence's case, hot air.

wayward is misquoting Pence. He didn't say its "only" increased testing that has caused the spike....
Pence said "And that in most of the cases where we are seeing some marginal rise in number, that’s more a result of the extraordinary work you’re doing.”

Do you think that quoted sentence is both true and consistent with administration policy?

Or is it only true if you disconnect it from policy?

https://news.trust.org/item/20200616122715-0bxw1

Talk about struggling to have it both ways.

which quoted sentence?
when wayward lawyerly confabulated the word "only" into pences statement he fubared its meaning.
who is struggling to have it both ways?
pence appeared to be congratulating health care workers for some
their efforts,

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Respond

Add Comment

Looks like good work, keeping in mind that Ellison is an economist who AFAIK hasn't worked with epidemiological models before. This line despite being a simple truism is something that people needed to keep in mind more, and still need to:

"Heterogeneous models have more parameters that need to be calibrated. Long run outcomes can be sensitive to activity levels of the less active, and it is difficult to calibrate these parameters early in an epidemic when there are few cases in less-active communities."

Even better is the next line:

"This is particularly true when one contemplates removing restrictions and thereby increasing activity among the currently inactive."

Yes, not only are people heterogeneous, in addition they can and do change their behavior, from more active to less active and vice versa. This endogeneity of behavior is what makes the social sciences much more complex than the natural sciences; molecules don't change their behavior (or if they do, they do so according to fixed rules).

Why might his work be useful? Because, if he's got it right,

"the models suggest fairly easy ways in which economists could extend their models and also point to data opportunities that might reduce the critical uncertainties"

That is, eventually, after the pandemic is over or has stabilized, we'll have a pretty good idea of what happened: how many people will get infected, how many will die, at what point did we reach herd immunity, etc. etc.

But early on (including right now) we do not know the values of the critical parameters. Worse, we do not even know what the critical parameters are, beyond highly generic ones such as Rt. His model, if it's sound, could guide us to figuring out what we need to be measuring and monitoring and what those values might mean.

So there's some good potential there. But is his model sound? I dunno; he's a good economist so there's that.

There's also the question about the quality of the data, we don't even have uniformity on deciding if Patient X died of COVID-19 or not. It's better to have some data than none at all but models or parameters that need fine-tuned accuracy are not going to get the needed accurate data. OTOH they could still be used to estimate how much uncertainty we still have, i.e. give plausible ranges of outcomes. And maybe Ellison's sorts of models will give ranges that are both more narrow and more plausible.

Respond

Add Comment

The key failure in the early models is that did not identify the data needed to estimate them.

the early modelers didn't have the required data because it is a new virus
and it was very early in the pandenic

Respond

Add Comment

That's not necessarily true. A number of early models that I read were spot on. There was a Greek/Italian paper that came out in early March that predicted the end of the Lomardy outbreak almost to the day. There were others as well.

Respond

Add Comment

Respond

Add Comment

This paper reaches conclusions similar to ones I worked on a couple of weeks ago.

https://gist.github.com/mike529/82ff801366fccc096f6bab930d3ed463

In particular even a small number of superspreaders (R0>>1) can seed a large outbreak even if the main population is not very contagious.

Respond

Add Comment

Respond

Add Comment