KEDRec-LM: A Knowledge-distilled Explainable Drug Recommendation Large Language Model

2025-02-27

AI Summary

This work addresses the challenge of interpretable drug recommendation. We propose KEDRec-LM, a large language model (LLM) trained via knowledge distillation–driven instruction fine-tuning, and introduce expRxRec—the first publicly available, multi-source heterogeneous dataset integrating knowledge graphs, clinical trial records, and PubMed literature. Methodologically, we pioneer the integration of knowledge distillation with instruction tuning, jointly leveraging drug graph embeddings, clinical text encodings, and PubMed semantic alignment to jointly generate accurate drug recommendations and natural-language medical explanations. Experimental results demonstrate that KEDRec-LM significantly outperforms existing baselines in both recommendation accuracy and rationale plausibility. Both the expRxRec dataset and the KEDRec-LM model are fully open-sourced, establishing a new benchmark and practical toolkit for interpretable biomedical AI research.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Other Foundations of Knowledge Representation & Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Drug discovery is a critical task in biomedical natural language processing (NLP), yet explainable drug discovery remains underexplored. Meanwhile, large language models (LLMs) have shown remarkable abilities in natural language understanding and generation. Leveraging LLMs for explainable drug discovery has the potential to improve downstream tasks and real-world applications. In this study, we utilize open-source drug knowledge graphs, clinical trial data, and PubMed publications to construct a comprehensive dataset for the explainable drug discovery task, named extbf{expRxRec}. Furthermore, we introduce extbf{KEDRec-LM}, an instruction-tuned LLM which distills knowledge from rich medical knowledge corpus for drug recommendation and rationale generation. To encourage further research in this area, we will publicly releasefootnote{A copy is attached with this submission} both the dataset and KEDRec-LM.

Problem

• Explainable drug discovery using LLMs
• Construction of comprehensive drug dataset
• Knowledge-distilled LLM for drug recommendation

Innovation

• Knowledge-distilled LLM for drug recommendation
• Instruction-tuned model for rationale generation
• Open-source dataset from medical knowledge graphs