Scholarlink

<p>With the widespread use of Artificial Intelligence (AI) in healthcare, how human physicians work with AI, a new colleague, has become an issue of increasing concern. This review systematically explores the collaboration between humans and AI in the medical field, focusing on the prevalent human-AI collaboration patterns, the task division mechanisms, and the evaluation metrics. Following the PRISMA guidelines for scoping reviews, we screened the relevant literature and finally included 85 journal papers. Our analysis reveals that human-AI collaboration patterns in healthcare tend to be associated with the inherent risk and automation level of medical tasks, and the collaboration process demonstrates a clear division of labor between human physicians and AI. Furthermore, the evaluation of the human-AI collaboration effect is shifting from single technical metrics (e.g., diagnostic accuracy) to multidimensional aspects incorporating clinical efficiency and user experience. Overall, this review presents collaboration patterns in various medical tasks and serves as a theoretical basis for optimizing the clinical workflows for human-AI collaboration.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A scoping review of human-AI collaboration patterns and task divisions in healthcare applications

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月19日

查看全文
  • Yuxuan You,
  • Xi Li

摘要

With the widespread use of Artificial Intelligence (AI) in healthcare, how human physicians work with AI, a new colleague, has become an issue of increasing concern. This review systematically explores the collaboration between humans and AI in the medical field, focusing on the prevalent human-AI collaboration patterns, the task division mechanisms, and the evaluation metrics. Following the PRISMA guidelines for scoping reviews, we screened the relevant literature and finally included 85 journal papers. Our analysis reveals that human-AI collaboration patterns in healthcare tend to be associated with the inherent risk and automation level of medical tasks, and the collaboration process demonstrates a clear division of labor between human physicians and AI. Furthermore, the evaluation of the human-AI collaboration effect is shifting from single technical metrics (e.g., diagnostic accuracy) to multidimensional aspects incorporating clinical efficiency and user experience. Overall, this review presents collaboration patterns in various medical tasks and serves as a theoretical basis for optimizing the clinical workflows for human-AI collaboration.