面向任务的多智能体框架用于复杂可穿戴健康分析
原标题：A Task-Oriented Multi-Agent Framework for Complex Wearable Health Analysis
AI 摘要
该论文提出一种面向任务的多智能体框架，用于处理复杂的可穿戴健康查询。框架通过意图-任务层级将复合查询拆解为不同意图和带显式依赖的类型化任务，由专门智能体分别执行检索、分析和建议任务，并保持各意图状态隔离。在含一个月纵向可穿戴记录的合成数据集上，Query Agent 在检索问题上准确率优于直接调用 LLM 的基线，并将查询阶段平均 token 消耗从较高值降至较低值；Manager Agent 在多意图问题上取得一定的多意图覆盖率和 Multiset Jaccard 相似度。该方法在可信度和透明度上得分更高，但可操作性未持续提升，健康建议生成与真实数据验证仍是开放挑战。
正文节选
A Task-Oriented Multi-Agent Framework for Complex Wearable Health Analysis Abstract Wearable health questions often combine data retrieval, longitudinal analysis, and health advice over structured records. Prompting a single large language model with a complete record and a composite query obscures whether every request is executed and which evidence supports the answer. We propose a task-oriented multi-agent framework that represents a composite query as distinct intents and typed tasks with ex