MARC v1: An Open-Source Multi-Agent Framework for Clinical AI Reasoning and Coordination· MARC v1: 开源多智能体临床 AI 推理与协调框架
We present Multi-Agent Reasoning and Coordination (MARC), an open-source framework that replaces monolithic LLM prompting with deterministic multi-agent orchestration for clinical reasoning. MARC coordinates role-specialized agents for extraction, reasoning, answer generation, and evaluation, with explicit context passing and traceable intermediate outputs, enabling stage-wise failure attribution. We additionally introduce a Decomposer module that generates task-specific agent prompts from a plain-language description, eliminating manual prompt engineering. The framework supports both API-based and local CPU-compatible deployments and is entirely configurable via YAML, without code modifications. MARC is designed to be model-agnostic, interpretable, and accessible to clinical domain expert
MARC v1 提供了一种多智能体协作的临床 AI 推理框架。
- 核心方法
- MARC 引入了一个确定性的多智能体协调框架，包括提取、推理、生成答案和评估等角色特化的智能体，并通过显式上下文传递和可追溯的中间输出来实现阶段性的错误归属。另外，还添加了一个将自然语言描述转换为特定任务提示的解构器模块，避免了手动提示工程。
- 适合谁读
- 研究者、工程师、临床专家
- 要解决的问题
- 传统的大规模语言模型（LLM）单一体提示方法在临床推理中存在局限性，难以追踪中间步骤和分配错误原因。
- 关键实验
- 未提供
- 主要贡献
- 提供了开源的多智能体临床 AI 推理框架，支持API和本地CPU部署，完全通过YAML配置，无需代码修改，模型无关且可解释。
- 意义与局限
- 提高了临床AI应用的灵活性和可解释性，为临床专家提供了更容易使用和理解的工具。然而，其实际临床效果和效率仍需进一步验证。