JMIR Publications

ABSTRACT

Objective:

To develop and evaluate an intelligent, automated protocol to identify and classify HPV vaccine misinformation on social media, using machine learning (ML)-based methods.

Methods:

Reddit posts (2007–2017, N = 28,121) were compiled that contained human papillomavirus (HPV) vaccine-related keywords. A two-step pipeline was proposed for misinformation identification and classification. A random subset (N = 2,200) was manually labeled for misinformation and served for the training and evaluation of ML algorithms (e.g., convolutional neural network [CNN]) for misinformation identification. The trained CNN model was applied to identify the misinformation from un-labeled posts. Then, for the posts that were inferred containing misinformation, topic modeling was further applied to identify the major categories (i.e., classification) associated with HPV vaccine misinformation.

Results:

The CNN model achieved the highest area under the receiver operating characteristic curve (AUC) at 0.7943 in the identification of misinformation. Of 28,121 Reddit posts, 7,207 (25.63%) were identified containing misinformation. Topic modeling then classified major misinformation categories from these posts, including general safety issues, which was identified as the leading type of misinformed posts (37%). Conclusion: ML-based approaches are effective in the identification and classification of HPV vaccine misinformation from Reddit and may be generalizable to other social media platforms. Public health implications: ML-based methods may provide the capacity and utility to meet the challenge of intelligent, automated monitoring and classification of public health misinformation in social media networks.