Authors:
Robert Elsässer
1
;
Adrian Ogierman
2
and
Michael Meier
2
Affiliations:
1
University of Salzburg, Austria
;
2
University of Paderborn, Germany
Keyword(s):
Epidemic Algorithms, Power Law Distribution, Disease Spreading, Public Health.
Related
Ontology
Subjects/Areas/Topics:
Agent Based Modeling and Simulation
;
Application Domains
;
Applications and Uses
;
Biological Systems
;
Biomedical Engineering
;
Complex Systems Modeling and Simulation
;
Environmental Modeling
;
Health Information Systems
;
Healthcare
;
Sensor Networks
;
Simulation and Modeling
;
Simulation Tools and Platforms
;
Social Systems Simulation
;
Software and Architectures
Abstract:
In a world where epidemic outbreaks may take many lives, forecasting and analysis tools are of high importance
- for an urban area such as New York City, a continent like Africa, as well as for the world itself. Such
tools provide valuable insight on different levels and help to establish and improve embankment mechanisms.
In this paper, we present an agent-based algorithmic framework to simulate the spread of epidemic diseases. Based on the population structure of Germany, we investigate the impact of the number of agents, representing the population, on the quality of the simulation. Real world data provided by the Robert Koch Institute (Arbeitsgemeinschaft Influenza, 2011; Robert Koch Institute, 2012) is used to evaluate our results. In a second step we empirically analyze the effects of certain non-pharmaceutical countermeasures as applied in the USA against the Influenza Pandemic in 1918-1919 (Markel et al., 2007). Our simulation and evaluation tool partially relies on the proba
bilistic movement model presented in (Elsässer and Ogierman, 2012). Based on our empirical tests, we conclude that the amount of agents in use can have a huge impact on the accuracy of the achieved simulation results. This reveals several challenges, which have to be taken into account in the design of forecasting and analysis tools for the spread of epidemics. On the other hand, we show that by utilizing the right parameters in our algorithmic framework - some of them being obtained from real world observations (Eubank et al., 2004) - one can efficiently approximate the course of a disease in real world.
(More)