Identifying False Human Papillomavirus (HPV) Vaccine Information and Corresponding Risk Perceptions From Twitter:...

Area of Science:

• Public Health
• Computational Linguistics
• Vaccinology

Background:

• Low human papillomavirus (HPV) vaccine uptake persists despite proven effectiveness.
• Vaccine hesitancy is exacerbated by the spread of misinformation on social media.
• Addressing hesitancy requires combating false HPV vaccine narratives online.

Purpose of the Study:

• To develop a systematic and generalizable method for identifying false HPV vaccine information on social media.
• To analyze the characteristics of true and false HPV vaccine information.
• To understand risk perceptions associated with HPV vaccines.

Main Methods:

• Utilized machine learning and natural language processing to analyze Twitter data.
• Developed classification models, including convolutional neural networks, to distinguish true from false information.
• Employed unsupervised causality mining to identify vaccine risk perceptions.

Main Results:

• A convolutional neural network model achieved high accuracy (F score=91.95) in identifying false HPV vaccine tweets.
• False information often used loss-framed messages and diverse vocabulary, focusing on potential risks.
• True information utilized both gain- and loss-framed messages, with a narrower vocabulary, emphasizing vaccine effectiveness.

Conclusions:

• Predictive models are feasible and effective for identifying false HPV vaccine information on social media.
• This methodology can help detect and analyze misinformation, contributing to public health efforts.
• Understanding risk perceptions associated with vaccine misinformation is crucial.