This app estimates the numbers of fully vaccinated, partially vaccinated, and unvaccinated people in a group at a future date. These people can be from different locations and from different age groups subject to data availability.
This app consists of the following steps:
Created by Bryn Wiley. Thank you to Dr. Sarah Otto, Dr. Daniel J. McDonald, and the BC COVID-19 Modelling Group for their substantial input and suggestions.
To add or remove group members either
Location: A location for individuals to originate from (eg: United States)
Region: A region of a selected location. Currently only supports US states or Canadian provinces
Age: An age group for individuals from the selected location and region. Not available for all locations or regions.
Number of People: The number of people from the selected location, region, and age group.
Maximum Portion Vaccinated: The maximum allowable portion of people vaccinated. If you have an upper limit on what you expect the maximum portion of people to be you can set this to a value between 0 and 1.
To ensure required group data is present, use this template for file upload.
This is the date on which vaccination status estimation will be made. It is not recommended to specify this more than 3 weeks in the future.
This specifies how the app will estimate future vaccination rates. There are two options:
Review the past and future estimated vaccination rates per location, region, and age group
Review the estimated numbers or percentages of people unvaccinated, partially vaccinated, or fully vaccinated in the group
Disclamer: Estimations of future vaccination rates are made based on available data, which can be infrequent or incomplete (see below). These predictions are made by extrapolating from recent trends, and changes in vaccination policy, vaccination availability, and anything that makes the individuals gathering less representative of the places they come from would cause errors in predictions. Projecting vaccinations into the future is overall highly uncertain, and is subject to increasing error the further the event is into the future.
Vaccination data is taken from Our World in Data, the US CDC, and the Public Health Agency of Canada. There are some countries who will primarily report total doses given instead of first or second doses. If a country has not reported a quantity for first or second doses within the past month we assume that the proportion fully vaccinated remains constant from the last date it was reported and all new reported doses are first doses. This is conservative with regards to fully vaccinated individuals but optimistic to the number partially vaccinated. Currently, the countries to which this applies are: Bonaire Sint Eustatius and Saba, Falkland Islands, Guernsey, Kuwait, Monaco, Nauru, Niue, Pitcairn, Qatar, Saint Helena, Sudan, Turkmenistan and Tuvalu
Prediction using linear regression
uses available data from the past 3 weeks to estimate a linear model with the
function in R.
If there are two or less observations in this period, regression is deemed not viable and so the vaccination rate is held constant.
Prediction using logistic regression
uses available data from the past 6 weeks to estimate a logistic (s-shaped) curve using the
function in R.
More recent observations are weighted stronger than observations further in the past. Starting values are generated using the
function, but if the starting estimate for the asymptote is greater than the maximum allowed value or less than the current vaccination rate,
it is adjusted to the maximum allowed value or the current value respectively. In the event that the
function fails to fit a logistic function, a linear regression is fit instead.
Again, if there are two or less observations in this period, regression is deemed not viable and so the vaccination rate is held constant
95% prediction intervals for estimated vaccination rates
are generated from the
function for linear regression, or from the
function from the
package in R for logistic regression.
If the upper bound of the prediction interval is larger than 1 or the maximum specified amount, it is reduced to this maximum amount. For the proportion fully vaccinated
this upper amount is either the maximum specified amount or the maximum predicted proportion with at least one vaccination. Similarly, if the lower bound of the prediction interval is smaller than the
most recent observation it is set to be equal to the most recent observation, as vaccination rates should not decrease.
Confidence intervals for total group vaccination percentages are estimated by taking the maximum and minimum predicted totals for each vaccination status, as generated by the vaccination rate prediction intervals from each category, and then adding a safe binomial confidence interval of 1/sqrt(N), where N is the total group size.
Centers for Disease Control and Prevention. (2021). COVID-19 Vaccinations in the United States,Jurisdiction [Data set]. Retrieved from https://data.cdc.gov/Vaccinations/COVID-19-Vaccinations-in-the-United-States-Jurisdi/unsk-b7fc
Mathieu, E., Ritchie, H., Ortiz-Ospina, E. et al. A global database of COVID-19 vaccinations. Nat Hum Behav (2021). https://doi.org/10.1038/s41562-021-01122-8
Public Health Agency of Canada. Canadian COVID-19 vaccination coverage report. Ottawa: Public Health Agency of Canada. https://health-infobase.canada.ca/covid-19/vaccination-coverage/
All relevant code is hosted at https://github.com/BrynWiley/GroupVaccinationStatus
Questions or comments? Email Bryn Wiley at firstname.lastname@example.org