Welcome to the COVID-19 simulation page of the PEAS Center. From March 24 on, we will publish simulation studies related to COVID-19. Note that this is work in progress. We are a modelling and simulation company. We developed a detailed System Dynamics simulation model related to COVID-19.
- We are experts in multi-scale systems modelling and simulation, not epidemiologists. We develop simulation models to help governments and NGOs to make decisions in the face of complexity and deep uncertainty. Although we are not epidemiologists, we nevertheless have a track record in modelling and simulating epidemics and outbreaks (including Ebola, Flu, A(H1N1), Zika, Lyme) under deep uncertainty and providing policy advice based on the results.
- This site contains Work In Progress (WIP) results. No rights can be derived from any of the results.
- We do not have a client or sponsor for this work. We would love to help. We are ready, willing, and able to provide advice to governments, veiligheidsregio’s, majors, NGOs. etc.
return to our site www.peas.center soon to look at
updates and reassessments. Our potential plan of action (if we have
sponsors or clients, we might change the order of tasks or change our
- To develop simulation results for multiple cohorts. These results are more suitable for assessing ICU capacity needs and hospital capacity needs, as well as fatalities.
- Simulate results for the Netherlands under deep uncertainty (both with our 1 population model and with our multi-cohort model)
- Simulate the situation in other countries under deep uncertainty (both with our 1 population model and with our multi-cohort model)
- Extend the model to include all countries in the world to assess the effects of countries in the international context.
- Extend the country model with the sub-national scales (provinces and municipalities) to provide help to majors, first for the Netherlands, possibly for other countries.
- These simulation runs are generated with the simplest version of our simulation model of COVID-19 in the Netherlands (without distinguishing between different age cohorts). Right after this study, we will adapt our multi-cohort version of the model (Youngsters, Adults, Elderly) to reality. Results with that multi-cohort version of the model will be better to (ensemble) forecast ICU needs, hospital needs, and fatalities.
- The model is not calibrated to data (yet). Results are nevertheless in line with what we are currently seeing in the data provided by the RIVM. The reason why we are not calibrating (yet), is that we want our glass box models to generate results that are close to what is happening in the real world for the right reasons, without over-fitting models to data. One of the reasons is that not all data is good (e.g., in the Netherlands, not everyone with COVID-19 symptoms is tested for COVID-19, which is why data with respect to positive testing cannot be used for calibration).
- We usually simulate models under deep uncertainty (i.e. across large uncertainty spaces). We still need to set up the uncertainty engine for COVID-19. Until then, we are only showing one simulation run, which is not the forecast of the future. It simply is 1 simulation run. When simulating models under deep uncertainty, many simulation runs are generated, and all reasoning happens on a large ensemble (thousands to even millions) of scenarios. Not a single one. Models simulated under deep uncertainty are useful for scenario generation and planning/policymaking in the face of uncertainty. Single model runs are illustrative at most.
- The reason we nevertheless are making these results available now is that we believe there is a strong need to understand what needs to be done (follow up the advice of the RIVM and other health authorities). Our plan of action to provide better and more detailed simulation runs is provided above.