---
id: 20260304-T0-16
title: "DIG to Heal实现多智能体协作决策"
title_en: "DIG to Heal Achieves Multi-Agent Collaborative Decision-Making"
url: https://ai.daily.yangsir.net/daily/20260304-T0-16
issue_date: 2026-03-04
publish_date: 2026-03-03T05:00:00.000Z
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2603.00309
---

# DIG to Heal实现多智能体协作决策

DIG to Heal 通过可解释动态决策路径，扩展通用智能体协作规模。该系统打破预设工作流限制，支持智能体动态分配角色与任务路径。在医疗诊断任务中，三个智能体协作准确率达 89.2%，较单智能体提升 21.5%，能生成可追溯的决策逻辑，便于人工审查与干预。

## English Version

**DIG to Heal Achieves Multi-Agent Collaborative Decision-Making**

DIG to Heal expands generalist agent collaboration scale through explainable dynamic decision paths. This system breaks predefined workflow limitations, supporting dynamic agent role and task path allocation. In medical diagnosis tasks, three-agent collaboration achieves 89.2% accuracy, a 21.5% improvement over single agents, and generates traceable decision logic for manual review and intervention.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2603.00309)

**详情页**：https://ai.daily.yangsir.net/daily/20260304-T0-16

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*