Fake Cures: User-centric Modeling of Health Misinformation in Social Media

TL;DR

This study develops a user-centric model to identify social media users who spread health misinformation, particularly about ineffective cancer treatments, achieving over 90% accuracy to aid public health interventions.

Contribution

It introduces a multi-stage user selection process and a classifier based on user attributes, writing style, and sentiment to detect misinformation propagators on Twitter.

Findings

Classifier accuracy exceeds 90% in identifying misinformation spreaders.

Features related to user attributes, writing style, and sentiment are effective indicators.

The model can assist public health efforts in targeting misinformation sources.

Abstract

Social media's unfettered access has made it an important venue for health discussion and a resource for patients and their loved ones. However, the quality of the information available, as well as the motivations of its posters, has been questioned. This work examines the individuals on social media that are posting questionable health-related information, and in particular promoting cancer treatments which have been shown to be ineffective (making it a kind of misinformation, willful or not). Using a multi-stage user selection process, we study 4,212 Twitter users who have posted about one of 139 such "treatments", and compare them to a baseline of users generally interested in cancer. Considering features capturing user attributes, writing style, and sentiment, we build a classifier which is able to identify users prone to propagate such misinformation at an accuracy of over 90%, providing a potential tool for public health officials to identify such individuals for preventive intervention.