After introducing Epidemiradar, the participatory platform that tracks respiratory infections in Spain, we now turn to the second tool of the EOSC-SIESTA Epidemiology use case. If Epidemiradar shows how an epidemic is unfolding, this tool helps anticipate how one might spread. The COVID-19 pandemic showed how much public health decisions depend on models that can make that kind of prediction. However, building a realistic epidemic model is a demanding task. It requires a detailed picture of the population: who lives in each household, how old they are, where they work or study, and how they move around every day. Putting that picture together means collecting, cleaning and combining large volumes of census, cadastral and mobility data before a single simulation can be run.
At the heart of the tool are synthetic populations for France, Spain and Italy, which are ready to use. They were built from official sources such as national census data, cadastral records and building databases. They recreate each country at household level. Every synthetic person has an age, a household, a home address and, where relevant, a workplace or school, with daily commutes that follow real mobility patterns. The result is a population that behaves statistically like the real one but is made entirely of virtual people. It serves as the common foundation for every simulation run on the platform. Because the methods are generic, they can be extended to other European countries.
Working with the tool follows a simple sequence. Users first choose between the two main families of models in computational epidemiology:
● Agent-based models follow individuals. Each synthetic person moves between home, work, school and other places following a realistic weekly routine, and infections can occur when people share the same place at the same time.
● Metapopulation models work with populations of geographical areas. They combine local transmission with the movement of people between areas.
Users then define the disease model, either by building their own or by starting from one of the preloaded models. In the Model Builder, this is done visually. Boxes represent health states such as susceptible, infected or recovered, and arrows represent the transitions between them, which together set out how the disease progresses.
Next comes the scenario. Users decide how many initial cases to introduce and where, how long the simulation should run, and which area to study, from a whole province down to a single district. The Simulation Manager lets them launch the run, follow its progress, and stop, restart or download the results.
Once a simulation is complete, the Visualization Engine shows the outbreak unfolding on an interactive map. Users can see the fraction of infected people in each area, broken down by disease state. They can also explore how incidence evolves over time and follow the prevalence curve for any region they select. This helps them spot hotspots, compare areas and understand how an epidemic might develop under different assumptions. Besides this, they can also download the results per disease state, municipality or census area in time locally for further systematic analyses.
Each user can see only their own simulations and results, which keeps work in progress confidential until researchers are ready to publish. The tool is designed for epidemiologists, epidemic modellers and public health professionals, but also for teachers and students who want to explore how diseases spread.