Abstract: Background　The quality of the initial response to acute health events is critical for disease prognosis, especially in pediatric emergencies. As "first responders" parents and other caregivers often face difficulties in judgment and emotional anxiety due to a lack of professional knowledge. While current WeChat-based consultation platforms hold empowering potential, existing research primarily focuses on training and equipment provision, lacking systematic exploration of real-time response mechanisms. Objective　To explore the empowerment mechanism of WeChat-based digital health platforms for first responders in pediatric acute health events, analyze their help-seeking characteristics, doctors' response strategies, and platform interaction structures, and provide a basis for optimizing remote emergency support systems. Methods　Five WeChat pediatric consultation groups established by Chengdu Angel Children's Hospital were selected. Chat records from 2024 were collected and, after de-identification, a corpus containing 347 independent consultation dialogues (totaling 5 546 messages) was constructed. Support Vector Machine (SVM) was used to classify dialogues related to acute symptoms, combined with Latent Dirichlet Allocation (LDA) topic modeling to identify high-frequency health problem topics. Conversation analysis and pragmatic pattern analysis were employed to perform phase annotation and node co-occurrence analysis on typical dialogues. Results　Six categories of acute health topics were identified: nighttime fever (32.6%), acute gastrointestinal reactions (21.4%), breathing difficulties (16.2%), consciousness abnormalities (11.8%), trauma and bumps (9.3%), and accidental medication ingestion (6.5%). Parents, as first responders, exhibited significant anxiety (urgent expressions in 88.6% of dialogues) and provided vague symptom descriptions, requiring an average of (3.2±1.1) rounds of follow-up questions for clarification. Doctors' responses exhibited a four-stage pragmatic pattern: "emotional assurance-structured guidance-judgment and direction-risk guidance". Node co-occurrence analysis revealed significant sequential associations between stages: the co-occurrence value between emotional reassurance and structured guidance was 0.82, and between structured guidance and judgment and advice was 0.76. Conclusion　WeChat group consultations have potential for immediate response and decision support in pediatric acute health events, but they face issues such as low information transfer efficiency and reliance on manual responses. Future efforts should enhance platform empowerment efficiency through intelligent Q&A guidance, AI pre-screening, and structured data output, promoting its transformation into an intelligent, systematic remote emergency sentinel system.

Key words: WeChat group consultations, First responder, Remote response mechanism, Sudden health incidents, Pediatrics

摘要： 背景　突发健康事件的初始响应质量对疾病预后具有关键影响，尤其在儿科急症中，作为“第一目击者”的家长或其他照护者因缺乏专业知识，常面临判断困难与情绪焦虑。当前微信群问诊平台虽具备赋能潜力，但现有研究多聚焦培训与设备配置，缺乏对实时响应机制的系统探讨。目的　探究基于微信群的数字健康平台在儿科突发健康事件中对第一目击者的赋能机制，分析其求助特征、医生响应策略及平台互动结构，为优化远程急救支持体系提供依据。方法　选取成都天使儿童医院建立的5个微信儿科问诊群，收集2024年的聊天记录，经去标识化处理后构建包含347段独立问诊对话（共5 546条消息）的语料库。采用支持向量机（SVM）进行突发症状相关对话分类，结合隐含狄利克雷分布（LDA）主题建模识别高频健康问题主题；运用会话分析与语用模式分析，对典型对话进行阶段标注与节点共现分析。结果　共识别出6类突发健康主题：夜间发热（32.6%）、急性胃肠道反应（21.4%）、呼吸困难（16.2%），意识异常（11.8%）、外伤与磕碰（9.3%）和药物误服（6.5%）。家长作为第一目击者表现出显著焦虑（88.6%对话含紧迫表达），症状描述模糊，平均需（3.2±1.1）轮追问澄清。医生响应呈现“情绪安抚-结构引导-判断建议-风险导诊”四阶段语用模式，节点共现分析显示各阶段间具有显著序列关联，情绪安抚与结构引导共现值为0.82，结构引导与判断建议的共现值为0.76。结论　微信群问诊在儿科突发健康事件中具有即时响应与决策支持潜力，但存在信息传递效率低、依赖人工响应等问题。未来应通过智能问答引导、AI预筛查与数据结构化输出等手段提升平台赋能效能，推动其向智能化、系统化远程急救前哨转型。

关键词: 微信群问诊, 第一目击者, 远程响应机制, 突发健康事件, 儿科