Maintain or Ease Corona Actions? ITWM Introduces New Decision Support Tool at Healthcare Hackathon

Interview With the Developers

Two urgent questions are currently concerning the population and, above all, decision-makers: How will the corona epidemic spread and what actions are appropriate and effective to contain it? In order to support local decision makers in planning their actions, three departments of the Fraunhofer ITWM are working on a decision support tool called EpideMSE as part of the anti-corona program of the Fraunhofer-Gesellschaft. In first screencasts of the simulator they demonstrate the features of the EpideMSE tool. Our experts Michael Helmling, Johanna Schneider and Neele Leithäuser answer questions about the software in an interview.

[Please note: The screencasts are only available in German.]

In the project EpideMSE you will work on a new tool for decision support during the corona pandemic. For which decisions should the tool provide support?

Michael Helmling: The EpideMSE tool provides information on the course of the epidemic, estimates of the number of unreported cases and forecasts of the further spread of the epidemic, which also take into account the effects of actions or their easing. The tool is designed to answer crucial questions such as

  • How dangerous is an increase in the reproduction rate for a certain region?
  • How low must it remain in order not to overburden the health care system?
  • How many tests must be carried out so that the spread of the infection does not escape in secret?


What kind of information do you provide exactly? How can the tool be imagined in practice?

Johanna Schneider: The tool consists of several parts: the spread of the epidemic, the estimated number of unreported cases, the prediction and the analysis of actions. We have created screencasts that give a first insight into the tool.

Every decision-maker needs an overview of the spread of the epidemic. The EpideMSE software allows you to colorize federal states or regions on the map according to infection rate, new infections or death rate. This gives people a quick overview of what is happening in Germany and in the regions. If the person selects a region, we provide further details in a table and the chronological charts.

A fundamental question is how many people are or were actually already infected with Covid-19. We estimate this dark figure for each region using statistical methods and on the basis of the data collected. You can see at a glance on the colored map how high the detection rate, i.e. the percentage of positively tested persons out of all infected persons, is in this region. also see the number of concealed infected persons per age group or over time can be seen.

Probably the most important question is: What will happen next? With the help of a simulation, we predict the epidemic's spread for the current situation. You can run the simulation not only throughout Germany, but also specifically for your federal state or district. We currently calculate the expected curve for infected persons, deaths and reproduction rates. However, more information such as the intensive care bed occupancy rate is to be added here.


In addition to the prediction based on the current situation, the development with various actions or easing is of course decisive. These can be freely configured in the EpideMSE software. It is important to know from when on which infection rate should be used in the model. As a result, we then compare the simulated spread with action with the prediction without action. The effects of actions can be estimated and compared in advance.

What data do you use for the simulations?

Michael Helmling: For the simulations, we use data from several sources: official data from the Robert Koch Institute at district level, updated daily via API retrieval from the NPGEO-Corona Hub 2020, population data from the Federal Statistical Office, population figures by age group from Eurostat, and the number of social contacts according to a model by Joël Mossong (Mossong et al. 2008). Data from Johns Hopkins University are also included, e.g. to identify time-independent parameters such as the mean infectious time.

Who is the new decision support tool aimed at?

Michael Helmling: First and foremost, we want to support political decision-makers at the local level, e.g. county councils and district administrators or health authorities who need local decision support to better assess the epidemic's spread for the citizens of their communities.


What is special about the EpideMSE tool?

Neele Leithäuser: On the one hand, historical data - as in many data portals - can be displayed geo-referenced. On the other hand, it is also possible to estimate the number of unreported cases on a local level. This is based on statistical missing data procedures, which are carried out by our experts in the Department of Mathematics for Vehicle Engineering. The aim is to investigate whether different mortality statistics that cannot be explained by biological factors can be adjusted by means of dark figures.

Michael Helmling: The SEIR-like prediction model developed by the colleagues in the Department Transport Processes allows us to estimate the further spread of the epidemic at the local level. It also shows the connection between actions and effects. 

Johanna Schneider: Of course, local authorities who want to cooperate with us can also express wishes. We are open to further ideas and would like to optimize the tool so that it provides decision-makers with exactly the information they need. Therefore we decided to present the prototype of the EpideMSE tool at the Healthcare Hackathon in Mainz in a workshop.

Let's take a look into the near future: What else is planned?

Johanna Schneider: We have already started to improve the prediction of the spread of Covid-19 in Germany. At the moment, the EpideMSE tool is used to simulate the epidemic for all people in one area together. In the next step, however, we want to additionally divide the population into age groups and take their different infection risks into account in the prediction. Furthermore, the effects of the epidemic course will be evaluated closer to the health care system, i.e. the prediction will also include the burden on intensive care units.

Michael Helmling: We also expect a significant improvement in the quality of prediction by linking regions. Perhaps a typical commuter city may have few new infections today, but tomorrow this could easily be different if new infections are brought by commuters. And this also applies vice versa, of course. Mobility is a crucial factor in the spread of infections and we want to incorporate this into the EpideMSE tool.


Finally, a look into the further future: In which areas could the tool's methods still be helpful?

Neele Leithäuser: Germany is currently in a phase in which we had to react quickly and decisively to the Corona crisis. However, researchers suspect that the SARS-CoV-2 virus will not disappear and that we must expect a second wave of infection in autumn. We want to support the preparations for combating the next wave of infection. To this end, we are developing mathematically strategies as to where and with which groups of people to test for corona in order to make the most efficient use of the available testing capacities. The same applies to the hoped-for case that a vaccine is developed. Here too, apart from the ethical question, the logistical question arises as to which groups of people should be vaccinated first and where. A question that can be answered very well with mathematics.