Social Media Representation and Illness Detection on Social Media: Foodborne Illness as a Case Study

Nina Cesare , Boston University
Quynh Nguyen, University of Maryland
Christan Grant, University of Oklahoma
Elaine Nsoesie, Boston University

Researchers widely recognize the importance of addressing bias in digital data but identifying and unpacking the nature of this bias is a work in progress. This study seeks to build on this literature assessing the utility of detecting foodborne illness on Twitter. It acknowledges that effective disease surveillance requires considering both who uses a specific platform, as well as how they use it. It first measures how the composition of users tweeting about foodborne illness symptoms differs from the composition of individuals impacted by foodborne illness. It then characterizes how individuals within specific demographic groups vary regarding how they discuss illness within this platform. Understanding the former will allow researchers to more accurately assess the association between digitally reported symptoms and offline illness, and the latter may improve the effectiveness and comprehensiveness of digital platforms as tools for identifying disease outbreak

See extended abstract

 Presented in Session 5. Health & Mortality 1