Seminar Details

Seminar Details


Nov 9

1:30 pm

Estimation of Social Contact Networks to Improve Influenza Simulation Models

Gail Potter

General Exam

University of Washington - Department of Statistics

Advisor: Mark Handcock

The current A (H1N1) influenza pandemic has posed questions to policymakers about the most effective interventions and resource mobilization strategies. Furthermore, mutation of the A (H5N1) \"avian\" influenza virus could also cause a pandemic with an estimated 60% case mortality rate in humans, requiring fast analysis of intervention and containment strategies. A stochastic simulation model can help determine the best strategies in case of a pandemic. Current policy is informed by results from three large-scale simulation models, one of which is operated by a collaboration of University of Washington and Los Alamos National Laboratory researchers. The models are based on Census and transportation data, and assume that people make random contacts within their households, classes and schools, and workgroups and workplaces. These model assumptions regarding social contact behavior have not been tested by analysis of social contact surveys. We will develop and apply social network methodology to analyze the POLYMOD survey, a diary of social contacts that respondents kept during a 24-hour period. A contact is defined to be a two-way conversation of at least three words and/or a physical contact, since these contacts are most important for influenza transmission. We have developed methods to infer within-household contact networks for this data set. We will next use exponential family random graph models to infer within-school contact networks and within-workplace contact networks. We will use these results to infer the entire contact network for a community of size 6,000. We will simulate disease transmission and intervention strategies over the inferred contact network, and compare the outcomes to simulations from the current UW/LANL simulation model applied to a community of size 6,000 with identical demographics. If outcomes are substantially different, the simulation model will be modified to include the network structures which impact transmission.