The design and use of an agent-based model to simulate the 1918 influenza epidemic at Norway House, Manitoba

Abstract

Agent-based modeling provides a new approach to the study of virgin soil epidemics like the 1918 flu. In this bottom-up simulation approach, a landscape can be created and populated with a heterogeneous group of agents who move and interact in ways that more closely resemble human behavior than is usually seen in other modeling techniques. In this project, an agent-based model was constructed to simulate the spread of the 1918 influenza pandemic through the Norway House community in Manitoba, Canada. Archival, ethnographic, epidemiological, and biological information were used to aid in designing the structure of the model and to estimate values for model parameters. During the epidemic, Norway House was a Hudson's Bay Company post and a Swampy Cree-Métis settlement with an economy based on hunting, fishing, and the fur trade. The community followed a traditional, seasonal travel pattern of summer aggregation and winter dispersal. The model was used to examine how seasonal community structures and associated population movement patterns may have influenced disease transmission and epidemic spread. Simulations of the model clearly demonstrate that human behavior can significantly influence epidemic outcomes.