Understanding Public Discourse on Bipolar Disorder: A Sentiment and Topic Modeling Analysis of Spanish-Language Tweets

Background/Objectives: Bipolar disorder (BD) is a chronic psychiatric condition associated with substantial emotional and psychosocial burden. Social media platforms provide an opportunity to examine how BD is discussed and emotionally framed outside clinical settings. Although Spanish-language public discourse on BD remains underexplored, this gap is particularly relevant given that linguistic and cultural factors may influence how mental health conditions are expressed, perceived, and communicated. Understanding these differences is essential for developing culturally sensitive public health strategies, improving mental health literacy, and addressing stigma in diverse populations. Methods: A retrospective observational study was conducted using Spanish-language tweets referring to BD posted between 2007 and 2023. After preprocessing, 170,411 unique tweets were analyzed. Latent Dirichlet Allocation (LDA) was applied to identify dominant thematic domains, and emotional tone was classified using a RoBERTa-based model adapted from Ekman’s basic emotions. Statistical analyses were performed to examine associations between topics, emotional categories, and the presence of clinical terminology. Results: Tweet activity related to BD remained low until 2019, followed by a marked increase peaking in 2020 and remaining elevated thereafter. Topic modeling identified seven main themes, predominantly centered on symptoms, illness-related life crises, social anxiety, help-seeking, and the impact of BD on romantic relationships, while diagnosis- and treatment-related topics were less frequent. Emotional analysis showed a clear predominance of sadness across all topics, accounting for approximately 75% of tweets. Anger, joy, and optimism were present at lower frequencies and varied across thematic domains. Clinical terminology was significantly more common in diagnosis- and treatment-related topics, whereas experiential and affective narratives dominated the remaining themes. Conclusions: Spanish-language discourse on BD on Twitter is largely characterized by emotional expression and lived experience rather than biomedical language. These findings highlight cultural relevance of social media as a complementary source for understanding public perceptions of BD and underscore the importance of culturally and linguistically specific analyses in digital mental health research.