MC 6460
Candidate
Brendon Phillips , Applied Mathematics, University of Waterloo
Title
Homophily, Synchrony and Spatial Autocorrelation as Signals of Critical Transition in Coupled Disease-Behaviour Systems
Abstract
We're bringing back the 50s! Vaccine-preventable diseases are all the rage these days, and lots of people are talking about it. With the current ease of creating and disseminating content through social networks and disreputable "news" sites, vaccine denialism is spreading. Access to these myths is now easier than ever, especially through social media. The ability for users to select their friends naturally produces echo chambers, which are groups that only contain users expressing the same beliefs.
Disease spread is efficiently controlled by herd immunity, where a sufficiently high proportion of the population in a region is vaccinated against that disease. As such, falling vaccination rates driven by anti-vaccination sentiment will lead to the loss of this herd immunity, and create opportunities for epidemics to wreak havoc in the population.
The close relationship between these social and disease processes suggests that patterns of social interaction may yield enough information to allow for the prediction of these outbreaks. In disease modelling, epidemics represent phase transitions; research has so far led to the discovery of relatively few early warning signals, which are characteristic behaviours that precede a critical transition.
The aim of this research project is to identify classes of early warning signals for coupled disease-behaviour systems. This is done by observing and collecting data from the clustering of like-minded users on social networks, and the corresponding disease spread on the contact network. Preliminary results have yielded three classes of these early warning signals, but we continue to examine the applicability and stability of the signals found so far, as well as searching for others.