MedPAO-Assist：可边缘部署的模态感知放射报告结构化智能体
原标题：MedPAO-Assist: Edge-Deployable, Modality-Aware Agentic AI for Radiology Report Structuring
AI 摘要
研究团队提出 MedPAO-Assist，一个可在边缘设备部署的多智能体框架，能从放射科医生的口述和语音指令直接生成结构化放射报告。系统基于 Qwen3-0.6B 小语言模型和任务专用 LoRA 适配器，采用 Extract-Ground-Generate 范式完成概念抽取、SNOMED-CT 本体对齐和报告生成，并支持语音驱动的人工修正。它仅需数 GB 内存，在普通笔记本 CPU 上端到端延迟为秒级，支持四种影像模态，全程无需键盘输入、外部 API 调用或联网，代码已开源。
正文节选
MedPAO-Assist: Edge-Deployable, Modality-Aware Agentic AI for Radiology Report Structuring Abstract The adoption of Large Language Models for radiology report structuring has been largely dominated by either proprietary cloud-based models or large open models requiring high-end GPUs for deployment, which is a critical bottleneck for clinical adoption in radiology workflows that demand data privacy and offline operation. To address this, we present MedPAO-Assist, an edge-deployable agentic framew