Mathematical epidemiologists have made strides towards building realistic and predictive models of disease transmission dynamics within human populations. These models are used to evaluate possible intervention strategies, that is, the demographic, spatial, and temporal distribution of costly measures and limited resources. However, the computational complexity of these models often precludes systematic optimization of such measures; one typical approach has been to compare the projected efficacies of a relatively small set of candidate strategies. Here, we present the use of the UCT algorithm for optimizing targeted intervention strategies. The algorithm searches large strategy spaces and can be adapted to work with existing epidemic simulators. We apply the optimization algorithm to a network model of disease transmission within and among North American cities, with the goal of optimizing antiviral control strategies for 2009 Pandemic H1N1. The disease model is populated with US Census Bureau and the Bureau of Transportation Statistics data and assumes disease progression and transmission parameters estimated for H1N1 North American influenza during the initial March-April 2009 outbreak in Mexico City. Our analysis yields time series of optimal interventions (amount of antivirals to release and method of release) under a variety of realistic scenarios.

You can also read the related paper on antiviral control strategies for influenza.

The following are two visualizations for the simulated spread of swine-origin H1N1 influenza.

Pictured above is the spread of H1N1 under several scenarios. The circle sizes are proportional to city population. When a circle is yellow, it means that enough antivirals are present to satisfy demand. When a circle turns gray, it means that there are not enough antivirals to satisfy demand. Inside of each circle is a smaller circle indicating the number of individuals who have been infected inside the city. When this inner circle is red, the epidemic is still spreading. When the inner circle is black, the effective reproductive ratio is less than 1, and the epidemic is dying. The blue ring around the cities represents the number of antivirals available in the city. The initial condition for all the scenarios pictured is 100,000 sick individuals distributed uniformly randomly across the United States, which gives rise to the synchronous spread of the disease across all cities. In the following visualization, we use a different initial condition.

Pictured above is the spread of H1N1 under an initial condition of 1000 sick individuals in Mexico City. The transportation network used for the model is populated using data from the U.S. Census Bureau and the Bureau of Transportation Statistics. It includes data for travel between U.S. cities, between U.S. and Mexican cities, but not between Mexican cities. The primary purpose of this visualization is to demonstrate the capabilities of a new visualization system we are developing with the help of the Texas Advanced Computing Center (TACC), in particular Greg Johnson. On the bottom left of the visualization are displayed SIR curves for the epidemic. Paths of disease transmission are highlighted in red as the visualization progresses. The user can zoom into different parts of the visualization, and even change to a more three dimensional earth projection.

**Warning**: Creating default object from empty value in **/home/nedd/neddimitrov.org/modules/CGExtensions/CGExtensions.module.php** on line **630**

## Add A Comment