CERSI Collaborators: Mark Dredze, PhD; Eric Leas, PhD, MPH; Paiheng Xu, MSE.

FDA Collaborators: Mario Navarro, PhD; Jonathan Kwan, MS; Jill Staton; Andie Malterud, M.S.; Stephanie Pitts, PhD.

CERSI Subcontractors: Health Watcher Inc., John W. Ayers, PhD, MA; Barnard College-Adam Poliak, PhD.

Project Start Date: May 2020

Regulatory Science Challenge

Effective tobacco control depends on a clear and up to date understanding of the tobacco product landscape. However, existing strategies rely mostly on cumbersome methods to discover and monitor tobacco products or cannot be cost-effectively scaled to yield timely insights. To fill this gap, we turn to social media. Monitoring social media greatly expands the scope of surveillance with millions of diverse participants describing their experiences with tobacco products any time on any day. Use of these data could substantially improve methods within tobacco regulatory science.

Project Description and Goals

Our goal is to advance tobacco product surveillance through the application of artificial intelligence (AI) methods. These are advanced statistical computer programs that perform tasks normally requiring a person. This goal will be fulfilled by two separate tasks. First, by analyzing millions of social media posts from Twitter and Reddit, we will discover what tobacco products are being discussed on social media by US consumers and how US consumers react to those products. In addition, using similar methods and context, we will discover perceived benefits and harms of tobacco products that social media users are discussing. The results of this work may help government agencies and public health advocates better understand consumer perceptions and the tobacco product landscape.