Simulating Sickness
Diseases spread like gossip through a social network. By studying these networks researchers can model and predict the path of diseases to better decide the most efficient way to prevent them. Traditionally these outbreaks are modeled using a combination of differential equations and uniform mixing assumptions, but a recent study conducted in Oregon found that dynamic bipartite graphs (similar to the ones we learned about in class) can be used to model contact patterns. These graphs are formed using data from the actual census, land-use, and population-mobility patterns. The article detailing their research efforts can be found at: http://www.nature.com/nature/journal/v429/n6988/full/nature02541.html. They claim that by strategically placing “disease sensors” at calculated network hubs it will no longer be necessary to vaccinate an entire population. By detecting diseases early and vaccinating certain sub-populations, they will be able to contain outbreaks in a more efficient and less-costly manner.
The research team uses simulation software called EpiSims to build a large-scale dynamic contact graph to model the population. This software builds a network constructed from census data containing age, gender, and income, along with other data about the transportation system. EpiSims creates an artificial population to model exposures to the disease. This social network allows the program to monitor the people’s individual exposure to a disease at a second-to-second basis. The network is a bipartite graph comprised of, “about 1.6 million vertices, with a giant component of about 1.5 million people and 180,000 locations.” This simulation software allows the researchers to project the “worst-case scenario” of the disease because the program does not run on real-time. With this so-called peak into the future, scientists and doctors can target the disease contact hubs and prevent spreading at the source. In addition to targeting densely exposed areas, the graph allows the researchers to target long-distance travelers because it takes traffic patterns into consideration. Theoretically this can stop a disease from spreading to different cities.
The analysis of social networks paired with simulation modeling, as opposed to differential equations, can potentially play a huge role in preventing disease outbreaks. This study that analyzed the social and traffic patterns of the population shows it is plausible to design a safe plan to decrease the total vaccinations while keeping residents disease-free.
-bd
Source: http://www.nature.com/nature/journal/v429/n6988/full/nature02541.html