Text Analysis of Spanish-Language TikTok Comments on Breastfeeding Using Large Language Models

Abstract

Background. Breastfeeding provides benefits for both mothers and infants, but its adoption and continuation may be affected by social, cultural, economic, and informational factors. In Spanish-speaking countries, such as those in Latin America, there are limited studies that discuss these factors. To complement these studies, this article provides a systematic study of breastfeeding using content generated by Spanish-speaking users of TikTok. Methods. A retrospective observational study was conducted based on the collection and analysis of public TikTok comments associated with breastfeeding-related videos published between 2021 and 2026. Using Large Language Models (LLMs), keywords, linguistic patterns, and recurring themes were identified to understand breastfeeding practices and opinions. Specifically, RoBERTuito was used to assign sentiments and emotions to each comment, and BERTopic was used to group comments into topics. Results. Of the 3,484 comments analyzed, 40.6% were negative, 36.4% neutral, and 23.0% positive, and frequent emotions were joy, sadness, and anger. Prominent topics involved milk supply formula and mixed feeding, reaction to the influence and others, and decision to breastfeed and rights. In these topics, negative comments were more predominant than positive and neutral comments. The topic analysis also revealed discussions about the role of men’s opinions in breastfeeding decisions. Conclusions. Spanish TikTok comments on breastfeeding involved mothers’ experiences, a polarized debate, and support for mothers. LLMs and topic modeling proved useful for analyzing these comments at scale. The findings could help design social media messages that address mothers’ barriers.