Assumptions management in simulation of infectious disease outbreaks

Abstract

Simulation of outbreaks of infectious disease is an important tool for understanding the dynamics of the outbreak process, the impact of disease and population properties, and the potential effect of interventions. However, the interpretation of the simulation results requires a clear understanding of the assumptions made in the underlying model. Typical simulation tasks, such as exploring the space of different scenarios for population and disease properties, require multiple runs with varying model parameters. For such complex tasks, the management of the assumptions made becomes a daunting and potentially error-prone undertaking. We report explicit assumptions management as an approach to capture, model, and document the assumptions for simulator runs. It was found possible to extend ontology-based simulation, which uses an ontological model to parameterize the simulator, to incorporate an assumptions model in the ontology. We conclude that explicit assumptions modeling should be part of any infectious disease simulation architecture from start.

Figure 1.

The layered structure of the simulator approach.

Figure 2.

Simulation architecture and workflow. The arrows and connectors illustrate the data flow and ontological relationships, respectively. It is possible to relate concepts in the models that make up the scenario ontology to the assumptions ontology. Furthermore, the scenario developer can relate the alternative scenarios to run (in the “array of scenarios”) to the assumptions ontology. The simulation engine takes as input the general scenario ontology, a set of simulation jobs with different scenario configurations, and the assumptions ontology.

Figure 3.

The assumptions model.