- Minor additional outputs
- Added skip.second.dose.fraction
- Added outputs for outcomes among vaccinated and unvaccinated
- Added parameters for relative transmission among age 0-11, relative probability of previously infected to be vaccinated
- Improved modelling of vaccinated and unvaccinated compartments to better match vaccine efficacy inputs
- When adding automatic interventions, do not consider dates of future interventions
- Update documentation
- Update
ProjectScenario
to quickly run new scenarios if interventions/vaccines/variants only change after last observed data
- New Excel input file format
- Accounts for vaccines, including dose-specific efficacy for blocking transmission and hospitalizations/deaths
- Accounts for age targeting in vaccine rollout
- Allows for variants, including the possibility of increased transmission, increased hospitalization/death, decreased vaccine efficacy, decreased duration of vaccine and/or natural immunity
- Fits to seroprevalence estimates, hospital admissions, cases (in addition to hospital census, ICU census, deaths)
- Runs quickly (~20 seconds - old LEMMA was taking over 12 hours)
- In order to run in reasonable time, we removed the calculation of credibility intervals by default and now only compute a point estimate. [in progress - bringing back credibility intervals as an option]
- Made len_inter an estimated parameter again
- Added an error term to the projections so that future projections reflected the degree to which past projections differed from past actuals. This should increase the width of the short-term credibility intervals so that around 95% of the actuals lie within the 95% interval.
- Make len_inter and frac_PUI fixed instead of estimated parameters (very minor impact on projections but increases speed)
- Rt plot shows credibility interval
- Moved credibility interval calculation to Stan based on the Santa Cruz County COVID-19 Model (https://github.com/jpmattern/seir-covid19). Many thanks to Jann Paul Mattern and Mikala Caton!
- Added fitting to ICU, deaths and cumulative hospital admissions
- CredibilityIntervalFromExcel now supports up to 9 interventions (see FAQ for details)
-
“Best Guess” was sometimes misinterpreted so we changed “Best Guess” to “User’s Prior Projection”. User’s Prior Projection is only displayed if niter < 100.
-
PDF now includes short-term (dates of observed data) and long-term (until Projection End Date) plots
-
plot.observed.data.long.term
,plot.observed.data.short.term
added to internal.args (Internal sheet). These logicals control plotting the observed data bounds on the long-term and short-term plots. -
lower.bound.label
,upper.bound.label
added to internal.args (Internal sheet). For example, in San Francisco these are "Confirmed COVID19" and "Confirmed COVID19 + 30%PUI".
- Case Studies was updated to include plots