JMIR Publications

ABSTRACT

This study uses state-of-the-art techniques from Natural Language processing to cluster groups of posts from Facebook into topics to gain a nuanced understanding of themes surrounding breast cancer posts in English and Spanish. Further, this study analyzed post likes, comments and shares to understand the user engagement relationship with breast cancer content. We used CrowdTangle’s API to collect posts between May 2023-2024 with a comprehensive set of hashtags and queries (N= 347,085). Using BERTopic and k-Means clustering, we generated topics in both languages which showed that 30 to 40% of posts were associated with awareness and breast cancer support. The metadata analysis showed that most engaging English content is often linked to leading cancer organizations such as Susan G. Komen, while the most engaged with Spanish-language content showed more entertainment, and political-oriented pages. For Spanish content, while information from local municipalities and entertainment pages may positively impact public health information from non-specialist sources may not always provide scientifically accurate or up-to-date information about breast cancer prevention, screening, and treatment. Thus, we recommend strong intervention for Spanish content to ensure accurate medical information to those who depend on social media for health information.