AI agents

This review evaluates AI agent systems built on large language models in healthcare, demonstrating improved clinical task accuracy over base models, especially in complex scenarios. Multi-agent architectures show promise but require further real-world validation.