COVIDHunter - Switzerland

An Accurate, Flexible, and Environment-Aware Open-Source COVID-19 Outbreak Simulation Model

A COVID-19 outbreak simulation model that evaluates the current mitigation measures (i.e., non-pharmaceutical intervention) that are applied to a region and provides insight into what strength the upcoming mitigation measure should be and for how long it should be applied, while considering the potential effect of environmental conditions. Our model accurately forecasts the numbers of cases, hospitalizations, and deaths for a given day. We use Switzerland as a use-case for all the experiments. However, our model is not limited to any specific region as the parameters of COVIDHunter are completely configurable. We explain our approach in detail and provide a comprehensive treatment of all datasets, models, and evaluation results with different model configurations here and here. COVIDHunter accurately forecasts for a given day:

  1. The reproduction number, R.
  2. The number of infected persons.
  3. The number of hospitalized persons.
  4. The number of deaths.
  5. The number of individuals at each stage of the COVID-19 infection (healthy, infected, contagious, and immune).
  6. The strength and the duration of each mitigation measure.

Other data: January-February 2021 | February-March 2021 | April-May 2021 | Nov 2021-Feb 2022
Other visualizations: January-February 2021 | February-March 2021 | April-May 2021 | Nov 2021-Feb 2022
To reproduce the exact same results for the Switzerland case study, please follow these instructions.

× Click and drag out a rectangle in the plot area to zoom in or use the full-screen mode from the menu at the top-right corner of the plot area

Mitigation Measure Strength

February 2020 - April 2021

Daily Reproduction Number, R

February 2020 - April 2021

Daily Number of COVID-19 Cases

February 2020 - April 2021

Daily Number of COVID-19 Hospitalizations

February 2020 - April 2021

Daily Number of COVID-19 Deaths

February 2020 - April 2021

Table of Abbreviations

Abbreviation Description Data Source
Observed R The reproduction number, R, reported officially by the Federal Office of Public Health (FOPH) of the Swiss Confederation. The R number describes how a pathogen spreads in a particular population by quantifying the average number of new infections caused by each infected person at a given point in time. Link
Expected Cases The expected true number of cases based on the number of hospitalizations, assuming that FOPH announces the true number of hospitalizations Link
Observed Hospitalizations The number of COVID-19 hospitalizations reported officially by the FOPH of the Swiss Confederation Link
Observed Excess Deaths The number of excess deaths, which we calculate as the difference between the observed number of deaths reported by the FOPH during 2020 and the expected number of deaths during 2020 (the average number of deaths for the last 5 years, 2015, 2016, 2017, 2018, and 2019). Link
ICL The corresponding number as calculated by the Imperial College London (ICL) model (https://mrc-ide.github.io/global-lmic-reports/). The prediction plot is for maintaining the same mitigation measures. Link
ICL+50% The corresponding predicted number as calculated by the ICL model when strengthening mitigation measures by 50%. Link
ICL-50% The corresponding predicted number as calculated by the ICL model when relaxing mitigation measures by 50%. Link
IBZ The corresponding number as calculated by the Theoretical Biology Group at ETH Zurich (IBZ) model (https://ibz-shiny.ethz.ch/covid-19-re-international/). Link
IHME The corresponding number as calculated by the Institute for Health Metrics and Evaluation (IHME) model (https://mrc-ide.github.io/global-lmic-reports/) Link
LSHTM The corresponding number as calculated by the London School of Hygiene & Tropical Medicine (LSHTM) model (https://cmmid.github.io/topics/covid19/global_cfr_estimates.html) Link
COVIDHunter The corresponding number as calculated by the COVIDHunter model when using three configurations: 1) CRW or CTC environmental change approach, 2) a certainty rate level of 50% or 100%, and 3) using mitigation measures with a strength of 0.7 or 0.3 on a scale from 0 to 1, where 1 refers to the strongest mitigation measure. We explain each of these terms in the following rows below. Link
CRW Harvard CRW (https://projects.iq.harvard.edu/covid19/home) environmental condition approach that considers both weather changes and air pollution.  
CTC CTC environmental condition approach that considers only weather changes. The CTC approach refers to our statistical analysis for the relationship between temperature and the number of COVID-19 cases in Switzerland. We find that for each 1 Celsius degree rise in daytime temperature, there is a 3.67% decrease in the daily number of confirmed cases.  
100% A certainty rate level of 100%, which means that FOPH reports 100% of the corresponding true number.  
50% A certainty rate level of 50%, which means that FOPH reports only 50% of the corresponding true number.  
M(t)=0.7 It means that we maintain the same strength of the currently applied mitigation measures. M(t)=0.7 means that the strength of the mitigation measures applied from 22 January to 22 February 2021 is 0.7 on a scale from 0 to 1, where 1 refers to the strongest mitigation measure.  
M(t)=0.35 The mitigation measures from 22 January to 22 February 2021 are relaxed by 50% compared to these mitigation measures that are applied right before 22 January 2021.
WithoutCTC_50% This plot represents the effect of excluding environmental changes from the COVIDHunter model, by setting Ce(t)=1 in Equation 1, which leads to an inaccurate evaluation of the mitigation measures. For example, during the summer of 2020 (between the two major waves of 2020), COVIDHunter (WithoutCTC_50%) evaluates the mitigation coefficient to be as high as 0.6. This means that the mitigation measures (only mandatory of wearing mask on public transport) applied during the summer of 2020 are only 14% more relaxed compared to the mitigation measures (e.g., closure of schools, restaurants, and borders, ban on small and large events) applied during the first wave, which is implausible. This highlights the importance of considering the effect of external environmental changes on simulating the spread of COVID-19. Unfortunately, environmental change effects are not considered by any of the IBZ, LSHTM, ICL, and IHME models, which we believe is a serious shortcoming of these prior models.