Contextual Biomedical Language Models for Imbalance-Aware Drug–Food Interaction Classification

This paper presents an imbalance-aware deep learning approach for Drug-Food Interaction (DFI) classification using BioBERT. The proposed method classifies interactions into Safe, Neutral, and Unsafe categories, achieving 85% accuracy and a macro F1-score of 0.77. The approach addresses class imbalance using Focal Loss and class-weighting strategies.