Inferring Social and Psychological Meaning in Social Media

Health surveillance is the first line in detecting an epidemic and understanding its causes and progression and preventing a pandemic. It generally consists of systematic collection and reporting of population-based data for the purposes of detecting, characterizing, and countering health threats. While this seems sound in theory, by the time patients present with symptoms, laboratory specimens are obtained, tested and confirmed for the organism; exposure to the general populace is most likely past the “tipping point”, potentially heading towards epidemic and pandemic levels. The population is at the mercy of the pathogen and its inherent (and mutating) characteristics. In the case of COVID-19 these characteristics are extremely troublesome, leading to the near exponential infection curves; the emphasis on “flattening the curve”; and the near draconian measures taken to counter the pandemic. Unfortunately, epidemiological surveillance, analysis, and forecasting has not kept pace in terms of leveraging information in the digital age. There is a potential wealth of information embedded within social networks that can be leveraged for better surveillance, earlier detection, and more accurate forecasting. In this effort, we propose to adapt and extend SURF’s classification and deep learning capability to analyze social networks to help combat COVID-19 and future disease spread.