[评论] 初级保健中的对话式诊断人工智能：它开口之后会发生什么？
摘要
患者在看临床医生之前，越来越多地使用对话式人工智能（AI），包括用于可能的诊断。然而，临床证据缺乏，且大多为模拟证据。例如，Articulate Medical Intelligence Explorer（AMIE）在文本咨询中优于初级保健医生，但其“患者”是标准化病人。1 一个例外是一项纳入2069例患者的随机试验，在该试验中，聊天机器人可生成病史、初步诊断和检查医嘱，有时无需工作人员支持，尽管其主要结局主要关注效率。
English
Patients increasingly use conversational artificial intelligence (AI), including for possible diagnoses, before seeing a clinician. However, clinical evidence is lacking and is mostly synthetic. The Articulate Medical Intelligence Explorer (AMIE), for example, outperformed primary care physicians in text consultations, but its patients were patient actors.1 One exception is a randomised trial of 2069 patients in which a chatbot generated histories, preliminary diagnoses, and test orders, sometimes without staff support, although its primary outcomes focused mainly on efficiency.