Scenario 3 Data

[ChrisRackauckas]

Hey, tracking this discussion here because there's a lot of moving parts. This is very raw, and we'll check things off as we find what we need, and modify the top post based on the changing landscape of what we need.

Case data:

• Expect time series data on I + R
• Start with an assumption on the recovery
• Possible additional: alternative measure for recovery rate
• Modeling assumption: use total infections from 2 weeks ago as R0, determine I0 and S0 from that
• Need time series for total population of US over time

Deaths and Hospitalizations

• Daily time series on number of patients admitted to the hospital all US
• time series for mortality
• 10 gig file on whether hospitalized or not => percentage for the difference in parameters
  • Plot the percentage over time by month, see if a constant assumption is okay or not,
  • If not, need to use the time series
• Any factor for underreporting estimate? Wastewater time series

Vaccinations

• Time series of vaccinations
• Hospitalization rate difference due to vaccination?
• Recovery rate difference due to vaccination?
• Mortality rate difference due to vaccination? Hospitalized and not hospitalized

Age-Stratification

• Previous data that is age stratified is cases, and hospitalizations
• 10 stratifications, by 10 years each
• Underreporting over time?
• Data for assumption on recovery rate with respect to age
• Aggregated contact matrix for beta over age, from Scenario 1

Reinfection

• Change in hospitalization for people who are reinfected
• State of new york, people who reinfected?
• Median time to reinfection
• It may require R -> S ===> R -> S2
• Maybe model recovered as vaccinated S?

[shivaram]

@ChrisRackauckas can you look at #26 ? And then check off things here if they match what is here?

[ChrisRackauckas]

@anandijain, @Vaibhavdixit02, @paulflang

[paulflang]

I cannot tick boxes, so commenting what I found:

• time series for mortality: 6-month-milestone/evaluation/scenario_3/ta_1/google-health-data/usa-cases-deaths.csv

[paulflang]

Case data:

• Expect time series data on I + R (6-month-milestone/evaluation/scenario_3/ta_1/google-health-data/usa-cases-hospitalized-by-age.csv) -> this is not I but contains tested, hospitalized and deceased, which can be used as proxy, I guess.
• Need time series for total population of US over time: that is similar to the above, point, isn't it? Not

[paulflang]

• Daily time series on number of patients admitted to the hospital all US, also in the last sheet I linked.

[paulflang]

• time series for mortality: this is actually also in the last sheet I linked. So perhaps a duplication, since it is already found in an issue above in a different sheet.

[paulflang]

Any factor for underreporting estimate? Wastewater time series: NOT FOUND

[paulflang]

• Time series of vaccinations: yes, there is a sheet called usa-varrinations.csv

[paulflang]

• Hospitalization rate difference due to vaccination?
• Recovery rate difference due to vaccination?
• Mortality rate difference due to vaccination? Hospitalized and not hospitalized
  • None of the above seem to be available per vaccination status.

[paulflang]

• Previous data that is age stratified is cases, and hospitalizations: yes
• 10 stratifications, by 10 years each: yes
• Underreporting over time?: no
• Data for assumption on recovery rate with respect to age: no
• Aggregated contact matrix for beta over age, from Scenario 1: no

[paulflang]

Change in hospitalization for people who are reinfected: no
State of new york, people who reinfected?: no
Median time to reinfection: no
It may require R -> S ===> R -> S2: Idk
Maybe model recovered as vaccinated S?: Idk

[shivaram]

Case data:

* Expect time series data on I + R (6-month-milestone/evaluation/scenario_3/ta_1/google-health-data/usa-cases-hospitalized-by-age.csv) -> this is not `I` but contains `tested`, `hospitalized` and `deceased`, which can be used as proxy, I guess.

* Need time series for total population of US over time: that is similar to the above, point, isn't it? Not

@paulflang I think the "new_confirmed" column refers to number of positive tests and hence infections (I). Schema is at https://github.com/GoogleCloudPlatform/covid-19-open-data/blob/main/docs/table-epidemiology.md

[shivaram]

@paulflang more data to compute rates at #32 -- for example this includes the answer to "10 gig file on whether hospitalized or not => percentage for the difference in parameters"

[paulflang]

@shivaram : can I and @ArnoStrouwen get the rights to tick off boxes in this issue?

[paulflang]

@shivaram : do you have wastewater data?

[shivaram]

We are working on wastewater or underreporting data now. @YohannParis has the power to give rights to tick boxes

[paulflang]

I've got the rights already. Thanks for working on underreporting/wastewater.

[shivaram]

@paulflang I added some undercount from seroprevalence the literature in #38

[paulflang]

Thanks. Is my interpretation correct, that is assumes zero reinfections (i.e. reinfections would not show up in changes in seroprevalence, but would show up in changes of case prevalence)?

[shivaram]

Thats a good question. My read on it is that not accounting for reinfections is a limitation of this study. They say

Several other limitations also may have led to an underestimation of seroprevalence, 
including the exclusion of specimens from people specifically seeking SARS-CoV-2 antibody testing, 
the inability of these sero-surveillance methods to detect reinfection (particularly during 
the Omicron phase[45], and the potential that some fully vaccinated people who are subsequently 
infected may develop levels of N-antibody that fall below the assay's limit of detection