Case studies and reports
Reports
- Paper: Characterizing the spread of CoViD-19, D. Karlen, July 13, 2020.
- introduces the pypm framework and describes methods to define comparitive statistics with weak model dependence and estimate their uncertainties
- Update: pyPM reference model 3.1, December 12, 2021.
- diagrams show the many infection cycles that are incorporated into the model for the variants of concern, along with independent susceptible populations representing breakthrough candidates and vaccine and natural immunity escape, and the vaccination and booster models.
- Paper: Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States
- compares the probabilistic accuracy of short-term forecasts submitted by 27 teams from April 2020 through October 2021. The pypm model is ranked in the top two.
Case studies
Click on the titles below to see detailed results from the studies.
January 22, 2023: 5 European nations
This study was prepared for inclusion in the European forecast-hub, coordinated by the European Centre for Disease Prevention and Control.
Four Omicron strains are included: BA.1, BA.2, BA.4/5, and BQ.1.
The analysis uses a new approach to incorporate population immunity, using the observed BA.2 wave to define the level of population immunity that existed at that time. Some escape from natural immunity is included for BA.4/5 or BQ.1. New variants have emerged during the resurgence in BA.5, in particular BQ.1*. This variant has a substantial growth advantage, but it appears that it does not have significant immunity evasion.
Some previous studies:
- November 20, 27
- October 9, 16, 23, 30
- September 4, 18
- August 21, 28
- July 17, 24, 31
- December 5
- November 7, 14, 14
- October 3, 10, 24, 31
- September 5, 12, 26
- August 15
January 22, 2023: USA by state
This report shows the result of the analysis of US state data, following the general approach first used in the July 17 study from Europe. The effect of BQ.1 has peaked, with almost all states having infection rates in rapid decline.
Some previous studies:
- December 4, 18
- November 27
- September 4
- July 10, 24
- May 15
- March 13
- February 6, 13, 20
- January 2, 9, 16, 23, 30
- December 5, 19, 26
- November 7, 14, 21
- October 3, 10, 24, 31
- September 5, 12, 26
- August 8, 15
- July 25
- April 18
- March 21
January 4, 2021: BC by health region
Omicron parameters are estimated from each Health Authority case data. Longer term projections are made for BC. Detailed studies are also shown, adjusting key parameters for the projections. Even considering the great deal of uncertainty, it appears that the rapid growth of omicron infections will likely produce hospitalization demands in excess of capacity.
Some previous studies:
- December 6, 16, 20
- November 10, 22
- October 4, 13, 25
- September 7, 13, 21, 28
- August 16, 24, 30
- July 26
- May 28
- March 9, 24
January 4, 2022: 8 provinces
Omicron is spreading rapidly around the globe.
Analyses of provincial case data are performed to estimate omicron parameters and longer term projections are shown. There is a great deal of uncertainty in these projections, but they demonstrate that the rapid growth in Omicron infections will likely overwhelm health care systems later in January and February.
Some previous studies:
- December 6, 16, 20, 29
- November 10, 22
- October 4, 13, 25
- September 7, 13, 21, 28
- August 18, 24, 30
- July 29
December 20, 2021: BC by health region
Cases are rising rapidly due to omicron. All regions, except Northern HA, have recent upward departures from previous trajectories. Omicron parameters are estimated from the HA case data.
Some previous studies:
- December 6, 16
- November 10, 22
- October 4, 13, 25
- September 7, 13, 21, 28
- August 16, 24, 30
- July 26
- May 28
- March 9, 24
July 4, 2021: 16 German states
This study was prepared for inclusion in the German forecast-hub, led by researchers at the Karlsruhe Institute of Technology and the Heidelberg Institute for Theoretical Studies.
The model fits include separate infection cycles for the B.1.1.7 and B.1.617.2 variants and takes into account the vaccination rates in each state.
May 28, 2021: BC by health region
Variants of concern are responsible for most cases in BC, and following the “circuit breaker”, all regions of BC saw case rates decline. Those restrictions were relaxed at the end of May.
This study includes, for the first time, the possible effects of B.1.617.2, a variant of concern that has a significant growth advantage over the currently dominant strain, B.1.1.7.
March 24, 2021: BC by health region
The B.1.1.7 variant is likely responsible for the recent growth in daily cases across BC.
The province does not make public the data necessary to estimate the critical parameters for B.1.1.7 in the province, so useful forecasts cannot be provided at this time.
This study illustrates possible levels of B.1.1.7 and growth advantage, to give an indication how quickly the situation could change.
March 9, 2021: BC by health region
The BC government makes available the number of cases each day, divided into sex, age, and health region. The data differs somewhat from that used in the Canada-wide studies, possibly due to corrections in the dates assigned to each positive case.
In October the province saw COVID-19 growth in all regions with the strongest growth in the two Vancouver health regons. At the beginning of November, new restrictions came into force in Vancouver, which reduced the rate of growth.
The Fraser Health region had see declining daily cases, since mid-November, unlike the other regions which had steady or growing case numbers. Starting in January, cases started growing again in the Fraser HA, at about 2 % per day. Only the Interior region is showing significant decline.
Forecasts for the next 4 weeks are provided. Comparisons of recent data and a forecast from last month are shown.
February 5, 2021: Israel during vaccine deployment
Israel currently has the highest rate of vaccinations per capita, with more than 40% of the population having received their first dose by early February. The case and vaccination time-series data from Israel are analyzed and amongst the older population, being nearly 90% vaccinated, the decline in cases is consistent with expectations from vaccination immunity.
This represents the first quantitative demonstration of the herd-effect to COVID-19 at a national scale.
November 22, 2020: USA by state
This is a analysis that uses data on hospitalizations recently made available by the US HHS. Prior to this analysis, hospitalization data from the Covid Tracking Project has been used.
October 5, 2020: California by age
California provides daily case, hospitalization, and death data by age group. This provides useful data to study how to model a non-homogenous population.
June 24, 2020: Brazil states
An initial study of Brazil data: fitting to death data only.