Suppr超能文献

ChatGPT 在医疗保健领域中的Prompt 工程路线图:观点研究。

A Road Map of Prompt Engineering for ChatGPT in Healthcare: A Perspective Study.

机构信息

School of Public Health Sciences, University of Waterloo, Waterloo, ON, Canada.

Department of Computer Science, University of Saskatchewan, Saskatoon, SK, Canada.

出版信息

Stud Health Technol Inform. 2024 Aug 22;316:998-1002. doi: 10.3233/SHTI240578.

Abstract

Generative AI models, such as ChatGPT, have significantly impacted healthcare through the strategic use of prompts to enhance precision, relevance, and ethical standards. This perspective explores the application of prompt engineering to tailor outputs specifically for healthcare stakeholders: patients, providers, policymakers, and researchers. A nine-stage process for prompt engineering in healthcare is proposed, encompassing identifying applications, understanding stakeholder needs, designing tailored prompts, iterative testing and refinement, ethical considerations, collaborative feedback, documentation, training, and continuous updates. A literature review focused on "Generative AI" or "ChatGPT," prompts, and healthcare informed this study, identifying key prompts through qualitative analysis and expert input. This systematic approach ensures that AI-generated prompts align with stakeholder requirements, offering valuable insights into symptoms, treatments, and prevention, thereby supporting informed decision-making among patients.

摘要

生成式 AI 模型,如 ChatGPT,通过战略性地使用提示来提高精准度、相关性和道德标准,对医疗保健产生了重大影响。本观点探讨了提示工程在为医疗保健利益相关者(患者、提供者、政策制定者和研究人员)定制输出方面的应用。提出了一个在医疗保健中进行提示工程的九阶段流程,包括确定应用、了解利益相关者需求、设计定制提示、迭代测试和改进、伦理考虑、协作反馈、文档记录、培训和持续更新。这项研究通过文献综述聚焦于“生成式 AI”或“ChatGPT”、提示和医疗保健,通过定性分析和专家意见确定了关键提示。这种系统方法确保 AI 生成的提示与利益相关者的需求保持一致,为症状、治疗和预防提供有价值的见解,从而支持患者做出明智的决策。

文献检索

告别复杂PubMed语法,用中文像聊天一样搜索,搜遍4000万医学文献。AI智能推荐,让科研检索更轻松。

立即免费搜索

文件翻译

保留排版,准确专业,支持PDF/Word/PPT等文件格式,支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述,25分钟生成高质量综述,智能提取关键信息,辅助科研写作。

立即免费体验