当前位置: X-MOL 学术Nat. Med. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Large language models could make natural language again the universal interface of healthcare
Nature Medicine ( IF 58.7 ) Pub Date : 2024-08-23 , DOI: 10.1038/s41591-024-03199-w
Jakob Nikolas Kather 1, 2, 3, 4 , Dyke Ferber 1, 2, 3 , Isabella C Wiest 3, 5 , Stephen Gilbert 3 , Daniel Truhn 6
Affiliation  

Human healthcare providers communicate in natural language, such as “a 25-year-old male, 2 days after laparoscopic appendectomy for acute appendicitis, with normal bowel movements, no fever, mobilized, and ready for discharge today”. Recorded in medical codes, this might read: “a patient with K35, having undergone CPT 44970, set for discharge”.

During their training, healthcare professionals are taught mnemonics to structure text using terms such as ‘airway, breathing, circulation, disability, exposure’ (ABCDE), ‘subjective, objective, assessment, plan’ (SOAP) or ‘symptoms, allergies, medications, past medical history, last meal, events leading up to the illness or injury’ (SAMPLE). Such mnemonics, which are in natural language, are meant to ensure that essential information is obtained from a patient. Mnemonics are loose structuring systems that are sufficient to transmit information and serve as the de facto standard in healthcare communication.



中文翻译:


大型语言模型可以使自然语言再次成为医疗保健的通用界面



人类医疗保健提供者以自然语言进行交流,例如“一名 25 岁的男性,腹腔镜阑尾切除术后 2 天因急性阑尾炎,排便正常,没有发烧,活动起来,今天准备出院”。记录在医疗代码中,这可能显示为:“一名 K35 患者,接受了 CPT 44970,准备出院”。


在培训期间,医疗保健专业人员学习助记符,以使用诸如“气道、呼吸、循环、残疾、暴露”(ABCDE)、“主观、客观、评估、计划”(SOAP) 或“症状、过敏、药物、既往病史、最后一餐、导致疾病或受伤的事件”(SAMPLE)等术语来构建文本。这种助记词采用自然语言,旨在确保从患者那里获得重要信息。助记词是松散的结构系统,足以传输信息并作为医疗保健通信的事实标准。

更新日期:2024-08-23
down
wechat
bug