印度母婴护理可审计紧急分诊系统
原标题：Auditable Emergency Triage for Maternal and Newborn Care in India
AI 摘要
Noora Health 在印度基于 WhatsApp 的母婴护理服务中，将原本由 LLM 直接分类紧急情况的 triage 系统重构为两步：LLM 按临床医生编写的词表抽取规范症状与患者背景，再由确定性规则引擎判断是否紧急。新系统将召回率从 0.565 提升至 0.810，F1 从 0.606 提升至 0.702，并带来可审计性，临床专家可逐阶段检查错误、独立添加规则而无需重新跑昂贵评估。部署后系统已分诊 152,421 条患者查询，标记 28,535 条（18.7%）为紧急，过度升级率 17.8%，未增加漏诊，临床医生已新增 48 条规则。
正文节选
Auditable Emergency Triage for Maternal and Newborn Care in India Abstract At Noora Health, our nurses answer more than 50,000 medical queries per month on our WhatsApp-based service that provides caregivers with on-demand support. Their most time-critical task is emergency triage: deciding which queries need immediate in-person attention. To support them, we built a system that uses a large language model (LLM) to classify whether a message is an emergency and provide a rationale for interpreta