Swarm Intelligence in Drug Discovery Applications: Unlocking Deeper Insights on the Identification and Optimization of Potential Drug Candidates.

Swarm-based analysis technology represents a class of computational approaches inspired by biological systems, such as bees or ants, to solve complex and high-dimensional problems through the collective behavior of interacting agents. This review provides an overview of swarm intelligence methods in drug discovery, covering foundational concepts, major algorithms, and representative applications in molecular docking, drug screening, de novo molecular design, and combinatorial chemical space exploration. We summarize classical swarm-based approaches and discuss recent hybrid frameworks integrating swarm intelligence with machine learning, deep learning, and large language model (LLM)-based multi-agent systems. In addition to highlighting their potential for adaptive search and multi-objective optimization, we critically examine current limitations, including scalability, convergence reliability, parameter sensitivity, and computational cost in high-dimensional biomedical settings. We further emphasize that many emerging frameworks, particularly LLM-enhanced and multi-agent swarm systems, remain at an early stage and have not yet been extensively validated in real-world drug discovery pipelines. Overall, swarm-based methods provide flexible and interpretable strategies for complex optimization tasks, while continued advances in data integration, benchmarking, and biologically informed modeling will be important for their broader application in drug discovery.