Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

adjusted prevalence calculations don't match the claimed formula #1149

Open
rmbellovin opened this issue Sep 25, 2021 · 3 comments · May be fixed by #1177
Open

adjusted prevalence calculations don't match the claimed formula #1149

rmbellovin opened this issue Sep 25, 2021 · 3 comments · May be fixed by #1177
Assignees
@beshaya
Labels
skill: backend engineering skill: model/research type: bug Something isn't working

Comments

@rmbellovin
Copy link

rmbellovin commented Sep 25, 2021 •
edited

Describe the bug
The src/posts/paper/7-basic-method.ts paper says that the formula for adjusting the prevalence ratio is prevalance_ratio = 1250 / (day_i + 25) * positive_test_rate ** 0.5 + 2 and points to https://covid19-projections.com/estimating-true-infections-revisited/ for justification.

The calculations in src/data/calculate.ts and /update_prevalence.py instead use (1000 / (day_i + 10)) * (positivityRate / 100) ** 0.5 + 2.

Neither of these matches the current formula at https://covid19-projections.com/estimating-true-infections-revisited/, which is prevalance_ratio = 1500 / (day_i + 50) * positive_test_rate ** 0.5 + 2

It's also unclear to me that the same prevalence ratio formula should be used for US and for countries with widespread asymptomatic lateral flow testing, but that's a separate question (and I don't have a recommendation for a replacement).
@rmbellovin rmbellovin added the type: bug Something isn't working label Sep 25, 2021
@congdonan
Copy link

I've also noticed for several states I've checked, that it's calculating vaccination percentages for a given area as the percentage of eligible population vaccinated as opposed to the percentage of the entire population vaccinated. Since kids under 12 can't be vaccinated yet, that makes the percentage of vaccinated people significantly lower in the general population than just calculating the percentage of people old enough to qualify that are vaccinated.

@blanchardjeremy
Copy link
Member

I've also noticed for several states I've checked, that it's calculating vaccination percentages for a given area as the percentage of eligible population vaccinated as opposed to the percentage of the entire population vaccinated. Since kids under 12 can't be vaccinated yet, that makes the percentage of vaccinated people significantly lower in the general population than just calculating the percentage of people old enough to qualify that are vaccinated.

Can you please open this as a separate issue? Thanks!

@beshaya
Copy link
Member

beshaya commented Oct 17, 2021

Hi @rmbellovin, thanks for catching the discrepancy. Note the update text on covid19-projections:

December 10, 2020 Update: We adjusted the constant in the prevalence ratio formula from a = 1500 / (day_i + 50) to a = 1000 / (day_i + 10). This results in a slightly higher prevalence ratio in the beginning of the pandemic and a lower prevalence ratio currently.

So the code currently matches their most up to date formula. I'll update the text in the white paper to match.

beshaya added a commit that referenced this issue Oct 17, 2021
@beshaya
Update whitepaper prevalence formula to match code
958cba1 
Fixes: #1149
@beshaya beshaya linked a pull request Oct 17, 2021 that will close this issue
Open
@beshaya beshaya self-assigned this Oct 17, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@beshaya beshaya

Labels
skill: backend engineering skill: model/research type: bug Something isn't working
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Update whitepaper prevalence formula to match code microCOVID/microCOVID
5 participants
@blanchardjeremy @oxytocinlove @beshaya @rmbellovin @congdonan