Language models for drug-drug interactions: current applications, pitfalls, and future directions

Key Points

• Language models show promise in automating drug-drug interactions (DDIs) extraction and prediction tasks, enhancing efficiency.
• Current applications can identify both known and novel DDIs, providing crucial support in drug safety assessments.
• Observational analysis reveals challenges in LLMs, including limitations in accuracy and reliability in identifying DDIs.
• Future directions emphasize improving AI frameworks to enhance DDI prediction and extraction capabilities.