![[ Visit Client Website ]](images/banner.gif)
Sunday, August 30, 2009
Grand Hall/Exhibit Hall
Background: We developed a computer program which integrated agent-based models of pertussis transmission and public health management. The program included an interface that allowed the user to interactively set policies or make individual cases decisions within a dynamic, stochastic setting and a logging system to record the decisions made. The overall goal of this project was to develop decision support tools that improved public health decision making and enhanced training in public health epidemiology
Methods: The agent-based model of Bordetella pertussis transmission comprised a synthetic community of ~14,000 individuals. Different types of scripted school outbreaks of pertussis were implemented to present challenges for decision-making. Data visualization tools available to users included epicurves, line lists, case totals, case investigation forms, and maps. Seven epidemiologists pilot tested the system by executing multiple runs. User’s expertise was determined two ways: by the Tier 2 Council of State and Territorial Epidemiologists Competency Form and by occupation data. Participants were given a post-simulation survey; questions were either a 7-point Likert scale or open ended. Log files and surveys were analyzed.
Results: Four novices and three experts in epidemiology completed 3-5 scenarios. 6000 discrete user interactions were recorded. Experts considered the ability to use simulations to explore aspects of prophylaxis and contact tracing the most useful feature because of the insight it provided about the impact of control decisions. Experts agreed the level of filtering and integration provided by our system was useful to explore data. Novices felt the simulation made their jobs easier and more efficient. They rated the maps and the epicurves the highest.
Conclusions: Usage varied across individuals with different levels of expertise in epidemiology. All seven epidemiologists rated the user interface highly for its navigability and esthetic appearance. A few enhancements were suggested with respect to training and display of data.
Methods: The agent-based model of Bordetella pertussis transmission comprised a synthetic community of ~14,000 individuals. Different types of scripted school outbreaks of pertussis were implemented to present challenges for decision-making. Data visualization tools available to users included epicurves, line lists, case totals, case investigation forms, and maps. Seven epidemiologists pilot tested the system by executing multiple runs. User’s expertise was determined two ways: by the Tier 2 Council of State and Territorial Epidemiologists Competency Form and by occupation data. Participants were given a post-simulation survey; questions were either a 7-point Likert scale or open ended. Log files and surveys were analyzed.
Results: Four novices and three experts in epidemiology completed 3-5 scenarios. 6000 discrete user interactions were recorded. Experts considered the ability to use simulations to explore aspects of prophylaxis and contact tracing the most useful feature because of the insight it provided about the impact of control decisions. Experts agreed the level of filtering and integration provided by our system was useful to explore data. Novices felt the simulation made their jobs easier and more efficient. They rated the maps and the epicurves the highest.
Conclusions: Usage varied across individuals with different levels of expertise in epidemiology. All seven epidemiologists rated the user interface highly for its navigability and esthetic appearance. A few enhancements were suggested with respect to training and display of data.