人工智能书写试验显示,医生文件记录时间下降了10%,在准确性和耗竭性方面的结果好坏参半。 An AI scribe trial showed a 10% drop in physician documentation time, with mixed results on accuracy and burnout.

发表在"新英格兰医学杂志"上的一项UCLA健康试验发现,人工智能书记,特别是Nabla,与对照组相比,平均每次访问医生文档处理时间减少了41秒 - - 约10% - - 微软DAX和Nabla与疲劳和工作负载的适度改善有关. A UCLA Health trial published in the New England Journal of Medicine found that AI scribes, particularly Nabla, reduced physician documentation time by an average of 41 seconds per visit—about 10%—compared to controls, with both Microsoft DAX and Nabla linked to modest improvements in burnout and workload. 该研究涉及14个专科的238名医生和两个月内72 000名病人的诊疗。 The study involved 238 physicians across 14 specialties and 72,000 patient visits over two months. 虽然大多数医生发现这些工具易于使用,有益于病人的参与,但大赦国际的笔记有时含有错误,包括遗漏和代名错误,还发生了一起轻微的安全事件。 While most physicians found the tools easy to use and beneficial for patient engagement, AI-generated notes occasionally contained errors, including omissions and pronoun mistakes, and one mild safety event occurred. 研究人员强调,人工智能书记需要积极的医生审查,并呼吁进行更长的多机构研究,以评估它们对护理质量,成本和结果的影响. Researchers stress that AI scribes require active physician review and call for longer, multi-institutional studies to assess their impact on care quality, costs, and outcomes.