On the Feasibility of Using Twitter Data to Assess the Global Circulation Patterns of Influenza Viruses

Abstract

Having the flu is something that everyone is familiar with, and the influenza season hits every year. The intensity and timing vary from year to year driven by climatic conditions and antigenic evolution, through mechanisms that are only partially understood. Most research agree that the virus originates in East and South-East (E-SE) Asia and spread throughout the world through human movement. In this thesis we explore the possibility of modelling this circulation pattern using a simple semi-stochastic mathematical model. Interestingly, this model exhibits chaotic behavior and is unable to confirm the above mentioned hypothesis. A separate approach is to analyze influenza incidence data. However, these data are subject to substantial underreporting (or complete lack of reporting) during the low-seasons. Some recent works have suggested using social media data to obtain proxies of influenza-like illness (ILI) data. In this thesis we discuss if it is possible to discern pattern or tendencies using data from Twitter. As the data used is collected only during a short time window, we can only say something about the feasibility of using this approach to analyze the global circulation of influenza viruses.