Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey of LLM-based Agents in Medicine: How far are we from Baymax?

Wenxuan Wang, Zizhan Ma, Zheng Wang, Chenghan Wu, Jiaming Ji, Wenting Chen, Xiang Li, Yixuan Yuan

TL;DR

This survey synthesizes the current state of LLM-based agents in medicine, detailing architectures, application domains, and evaluation frameworks. It presents a taxonomy of system profiles, planning methods, reasoning strategies, and external capacity enhancements, alongside concrete medical applications in decision support, data analytics, training, and service optimization. The paper also highlights key challenges—hallucination, multimodal integration, cross-department interoperability, and ethical/privacy concerns—and proposes directions such as reinforcement-learning-based medical reasoning, integration with physical systems, and enhanced simulation-driven training. Collectively, it provides a structured blueprint for researchers and practitioners aiming to advance reliable, safe, and clinically integrated LLM-based medical agents.

Abstract

Large Language Models (LLMs) are transforming healthcare through the development of LLM-based agents that can understand, reason about, and assist with medical tasks. This survey provides a comprehensive review of LLM-based agents in medicine, examining their architectures, applications, and challenges. We analyze the key components of medical agent systems, including system profiles, clinical planning mechanisms, medical reasoning frameworks, and external capacity enhancement. The survey covers major application scenarios such as clinical decision support, medical documentation, training simulations, and healthcare service optimization. We discuss evaluation frameworks and metrics used to assess these agents' performance in healthcare settings. While LLM-based agents show promise in enhancing healthcare delivery, several challenges remain, including hallucination management, multimodal integration, implementation barriers, and ethical considerations. The survey concludes by highlighting future research directions, including advances in medical reasoning inspired by recent developments in LLM architectures, integration with physical systems, and improvements in training simulations. This work provides researchers and practitioners with a structured overview of the current state and future prospects of LLM-based agents in medicine.

A Survey of LLM-based Agents in Medicine: How far are we from Baymax?

TL;DR

This survey synthesizes the current state of LLM-based agents in medicine, detailing architectures, application domains, and evaluation frameworks. It presents a taxonomy of system profiles, planning methods, reasoning strategies, and external capacity enhancements, alongside concrete medical applications in decision support, data analytics, training, and service optimization. The paper also highlights key challenges—hallucination, multimodal integration, cross-department interoperability, and ethical/privacy concerns—and proposes directions such as reinforcement-learning-based medical reasoning, integration with physical systems, and enhanced simulation-driven training. Collectively, it provides a structured blueprint for researchers and practitioners aiming to advance reliable, safe, and clinically integrated LLM-based medical agents.

Abstract

Large Language Models (LLMs) are transforming healthcare through the development of LLM-based agents that can understand, reason about, and assist with medical tasks. This survey provides a comprehensive review of LLM-based agents in medicine, examining their architectures, applications, and challenges. We analyze the key components of medical agent systems, including system profiles, clinical planning mechanisms, medical reasoning frameworks, and external capacity enhancement. The survey covers major application scenarios such as clinical decision support, medical documentation, training simulations, and healthcare service optimization. We discuss evaluation frameworks and metrics used to assess these agents' performance in healthcare settings. While LLM-based agents show promise in enhancing healthcare delivery, several challenges remain, including hallucination management, multimodal integration, implementation barriers, and ethical considerations. The survey concludes by highlighting future research directions, including advances in medical reasoning inspired by recent developments in LLM architectures, integration with physical systems, and improvements in training simulations. This work provides researchers and practitioners with a structured overview of the current state and future prospects of LLM-based agents in medicine.

Paper Structure

This paper contains 39 sections, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. LLM vs. LLM-based Agent
  4. Unique Considerations for LLM-based Agents in Medicine
  5. LLM-based Medical Agent Architecture
  6. Profile
  7. Clinical Planning
  8. Clinical Planning
  9. Medical Reasoning
  10. External Capacity Enhancement
  11. Application Scenarios
  12. Clinical Decision Support and Diagnosis
  13. Clinical Data Analytics and Documentation
  14. Medical Training and Simulation
  15. Healthcare Service Optimization
  16. ...and 24 more sections

Figures (1)

  • Figure 1: Conceptual framework of LLM-based medical agents. This figure depicts the architecture of the proposed LLM-based Medical Agent, consisting of system profile, external capacity enhancement, clinical palnning and medical reasoning. It supports four agent paradigms: a) Single Agent, b) Sequential Task Chain, c) Collaborative Experts, and d) Iterative Evolution. The framework integrates external tools and reasoning mechanisms to enable applications in medicine.