Identifying the conditions under which a specific drug has therapeutic effects on a target disease is crucial for clinical decision-making support. However, most existing biomedical information extraction methods have focused on the relationships between drugs and diseases, largely overlooking the context-specific conditions where such relations apply.
To address this issue, we introduce the task of extracting applicability conditions for therapeutic drug-disease relations from biomedical research literature.
We create the first dataset with manually annotated triples of drugs, diseases, and applicability conditions, comprising 1,119 drug-disease pairs. Using this dataset, we systematically evaluate the performance of various existing methods. Additionally, we propose a new method that enhances LoRA to consider drug-disease relationships. Our method consistently outperforms strong baselines across different evaluation settings.
The source code and dataset can be accessed at: GitHub Repository.
Blogger's Review: This study addresses a significant gap in drug-disease relation extraction by introducing the concept of applicability conditions, providing more targeted support for clinical applications. The proposed method is a promising step forward in advancing biomedical text mining research.