JCSE, vol. 19, no. 1, pp.1-15, 2025

DOI: http://dx.doi.org/10.5626/JCSE.2025.19.1.1

Survey on AI-Drug Discovery with Knowledge Graphs: Data, Algorithm, and Application

Daeun Kong, Yoojin Ha, HaEun Yoo, Dongmin Bang, and Sun Kim
Interdisciplinary Program in Artificial Intelligence, Seoul National University, Seoul, Korea

Abstract: Drug discovery is a complex, costly, and high-risk endeavor reliant on extensive experiments and clinical trials. Recent advances in artificial intelligence (AI) are transforming biomedical research by modeling complex biological relationships and accelerating therapeutic discovery. Central to these innovations are biomedical knowledge graphs (KGs), which systematically integrate diverse, heterogeneous data, from molecular interactions and genetic profiles to drug-disease associations. In particular, heterogeneous knowledge graphs (HKGs) capture complex biological phenomena through interconnected multi-modal data sources. This survey provides a comprehensive overview of AI-driven drug discovery via HKGs, detailing their definitions, construction methodologies, and evaluation criteria. We further review state-ofthe-art AI algorithms from graph representation learning to hybrid reasoning approaches, and examine their applications in key drug discovery tasks such as drug-target identification, drug repurposing, combination therapies, and integration with large language models. Through this investigation, we highlight emerging opportunities and future directions that ide researchers in harnessing the full potential of KGs for novel therapeutic development.

Keyword: No keyword

Full Paper:   8 Downloads, 32 View