Visualising Patterns Associated with Adverse Drug Reactions in French Forums

Area of Science:

• Pharmacovigilance and Social Media Analytics
• Computational Linguistics
• Drug Safety Surveillance

Background:

• Social media platforms offer a rich, albeit unstructured, source of real-world data for pharmacovigilance.
• Traditional pharmacovigilance methods may not fully leverage the insights available in online user-generated content.
• Identifying adverse drug reactions (ADRs) from large volumes of text data presents a significant challenge.

Purpose of the Study:

• To develop and implement a novel visualization method for pharmacovigilance specialists.
• To apply this method to analyze French discussion forums for ADR detection.
• To enhance the identification of patterns associated with adverse drug reactions using natural language processing.

Main Methods:

• Trained a word embedding model on posts from French discussion forums.
• Developed a novel visualization technique tailored for pharmacovigilance data.
• Applied the word embedding model to identify linguistic patterns indicative of ADRs.

Main Results:

• The implemented visualization method effectively aids pharmacovigilance specialists.
• The word embedding model facilitated the identification of patterns related to adverse drug reactions.
• The study demonstrates the utility of social media data in pharmacovigilance.

Conclusions:

• Novel visualization methods combined with natural language processing can enhance ADR detection from social media.
• French discussion forums serve as a valuable data source for pharmacovigilance.
• This approach offers a promising tool for proactive drug safety monitoring.