search

ImperialCollegeLondon / covid19model

Code for modelling estimated deaths and cases for COVID19.
MIT License
944 stars 271 forks source link

Add additional country - Finland #56

Open MansMeg opened 4 years ago

MansMeg commented 4 years ago

Is your feature request related to a problem? Please describe. We are currently working on including Finnish data in the model. We would be happy to contribute this to your repo. How should we do that in the best way?

Describe the solution you'd like I would need information on what files (csv) needs to be updated and how. ECDC data already have death counts for Finland, so that is already handled.

I guess the following files need to be updated:

  1. interventions.csv when interventions have been done.
  2. popt_ifr.csv: I'm not totally sure about this. Is this computed based on age groups outside the repo, since the file contain one value per country?
  3. ages.csv. This is not superclear to me. Is that the age groups sizes in 1000s?

With kind regards Måns

elmokki commented 4 years ago

I added Finland to the version 1 of the model in a quick and dirty way last week. Based on that:

  1. Yes. The report and the web page can be helpful for deciding what constitutes as which intervention. We're still debating what counts as the lockdown date for Finland though.

  2. The model uses on IFR per country. The IFR is calculated using data that is not in the repository. There's some documentation on how the calculation is done in the report: It's not empirical IFR from the countries, but rather based on a study estimating true IFR of the disease, weighted by both age distribution and some sort of a contact matrix. Last week I used Swedish IFR on my version 1 calculations as a placeholder, because I doubt the Finnish IFR is heavily different than that of other Nordics.

  3. It is in thousands as far as I know.

For version 1, there were also some minor code changes required for it to support 12 countries.

I'll send a mail to your assumed (Finnish) work e-mail address from my work e-mail!

MansMeg commented 4 years ago

Do so! See mansmagnusson.com for contact info.

s-mishra commented 4 years ago

Hi @MansMeg and @elmokki, good job. We are currently not in a position to independently add countries. For countries to be added we need to make sure all of our assumptions and data match. I am happy for you guys to work independently but work for a new country can't be just added because we also want to make sure IFR and age distribution we use are correct and verified.

MansMeg commented 4 years ago

Ok, so how do we help you to do this verification. We are working with this from Finland so we could do the verification you need, I suppose?

Thats why I had these questions above.

s-mishra commented 4 years ago

It is mainly working with the people who have better understanding of estimating IFR from contact matrices and other is how much data is really there to estimate with confidence.

elmokki commented 4 years ago

Would you happen to have a reference for estimating IFR from contact matrices? I'm very interested in getting the best possible estimates regardless of whether they'd be added here, and that's the part there wasn't code or thorough documentation for.

s-mishra commented 4 years ago

Hi @elmokki as it is said in the report we are using verity.et.al as our reference for IFR and contact matrices. Just that when the paper doesn't really have it for a country, we talk to them and work with them to get it correct; as they understand it more than we do.

elmokki commented 4 years ago

@s-mishra thanks! I seem to have missed that reference. I'll look into it!

MansMeg commented 4 years ago

I do think I can actually do that estimation, if you are interested. Or are you in contact with others in Finland currently?

Although, I have no intention to push this if you are not interested. It was just that it was explicitly stated that you welcomed additional countries as contributions.

MansMeg commented 4 years ago

Sorry, I missed the last comment. So if I understand it correctly, the popt_ifr file contains values of ifr from the verity et al paper. If they dont have the estimate, they need to supply it to you for you to be able to include another country?

s-mishra commented 4 years ago

Hi @MansMeg, sorry if I sounded rude. I am just saying even if you do the update we might not be able to get it in the main branch without going through it in detail. However, if everything works we will add. I just need to see Finland data to be sure we can act on it.

MansMeg commented 4 years ago

No problem, I fully understand that you would not blindly accept a PR from random people.

So, as mentioned, we are multiple people that want to include this for Finland. I suggest we do that and open a PR. Then you can look at it when you have time.

For us to add Finland for our sake, we need to know:

  1. popt_ifr.csv: This is the ifr from the verity.et.al paper? You are in contact with authors to get these values computed for all countries? Hence you do not want us to supply these data in a PR?
  2. ages.csv. This is age groups sizes in 1000s?

In the PR:

  1. Is there anything you need to simplify checking of our data?