Areas of interest and sentiment analysis towards second generation antipsychotics, lithium and mood stabilizing anticonvulsants: Unsupervised analysis using Twitter

Abstract

Background: Severe mental disorders like Schizophrenia and related psychotic disorders (SRD) or Bipolar Disorder (BD) require pharmacological treatment for relapse prevention and quality of life improvement. Yet, treatment adherence is a challenge, partly due to patients' attitudes and beliefs towards their medication. Social media listening offers insights into patient experiences and preferences, particularly in severe mental disorders.

Methods: All tweets posted between 2008 and 2022 mentioning the names of the main drugs used in SRD and BD were analyzed using advanced artificial intelligence techniques such as machine learning, and deep learning, along with natural language processing.

Results: In this 15-year study analyzing 893,289 tweets, second generation antipsychotics received more mentions in English tweets, whereas mood stabilizers received more tweets in Spanish. English tweets about economic and legal aspects displayed negative emotions, while Spanish tweets seeking advice showed surprise. Moreover, a recurring theme in Spanish tweets was the shortage of medications, evoking feelings of anger among users.

Limitations: This study's analysis of Twitter data, while insightful, may not fully capture the nuances of discussions due to the platform's brevity. Additionally, the wide therapeutic use of the studied drugs, complicates the isolation of disorder-specific discourse. Only English and Spanish tweets were examined, limiting the cultural breadth of the findings.

Conclusion: This study emphasizes the importance of social media research in understanding user perceptions of SRD and BD treatments. The results provide valuable insights for clinicians when considering how patients and the general public view and communicate about these treatments in the digital environment.

Keywords: Bipolar disorder; Machine learning; Natural language processing; Psychosis; Social media; Twitter.