This research demonstrates how Large Language Models can effectively predict drug-drug interactions by processing molecular structures, target organisms, and gene interaction data as text.

• Successfully applied various LLMs to predict potentially harmful drug combinations
• Processed complex pharmaceutical data in raw text format without specialized encodings
• Compared performance across different model types and configurations
• Established foundation for integrating LLMs into clinical decision support systems

Why it matters: Predicting drug interactions is crucial for patient safety and effective treatment, especially as polypharmacy increases. This approach could significantly accelerate the identification of risky drug combinations before clinical trials.

LLMs for Drug-Drug Interaction Prediction: A Comprehensive Comparison