JMIR Publications

ABSTRACT

Background:

The COVID-19 pandemic has been accompanied by an information epidemic or “infodemic”: too much information including false or misleading information in digital and physical environments during an acute public health event, which leads to confusion, risk-taking and behaviors that can harm health, and lead to mistrust in health authorities and public health response. The analytical method described is part of the WHO work to develop tools for an evidence-based response to the infodemic, enabling prioritization of health response activities.

Objective:

The aim of this work was to develop a practical, structured approach to identifying narratives in public online conversations on social media platforms where concerns or confusion exist or where narratives are gaining traction, and to provide actionable data to help WHO prioritize its risk communications efforts where it is most critical in addressing the COVID-19 infodemic.

Methods:

We developed a taxonomy to filter global COVID-19 public online conversations in social media content in English and French into five themes, with 35 sub themes. The taxonomy and its implementation were validated for retrieval precision and retrieval recall, and reviewed and adapted as the linguistic expression about the pandemic in online conversations changed over time. The aggregated data were analyzed for each sub themes by volume, velocity and the presence of questions, on a weekly basis, to detect signals of information voids where there was potential for confusion or for mis- or dis-information to thrive. A human analyst reviewed the themes for potential information voids and used quantitative data to provide context and insight on narratives, influencers and public reactions.

Results:

A COVID-19 public health social listening taxonomy was developed and applied. A weekly analysis of public online conversations since 23 March 2020 has enabled the quantification of shifts of public interest in public health-related topics concerning the pandemic and has demonstrated the frequent resumption of information voids with verified health information. This approach therefore focuses on infodemic signal detection for actionable intelligence to rapidly inform decision-making for a more effective response, including adapting risk communication.

Conclusions:

This approach been successfully applied during the COVID-19 pandemic to identify and take action on information voids based on analysis of infodemic signals. More broadly, the results have demonstrated the importance of ongoing monitoring and analysis of public online conversations, as information voids frequently resume and narratives shift over time. The approach is already being piloted in individual countries and WHO regions to generate localized insights and actions, while a pilot of an AI social listening platform is using this taxonomy to aggregate and compare online conversations across 20 countries. Looking beyond the COVID-19 pandemic, the taxonomy and methodology have the potential to be adapted for fast deployment in future public health events.