文献速递胃肠外科围术期管理与 ERAS

机器人手术中任务级自主性自动评估

Automated task-level autonomy assessment in robotic surgery.
PubMed 文献 · Surg Endosc · 2024-08-07

本研究回顾性分析机器人辅助食管裂孔疝修补术的手术表现数据,通过AI辅助标注将手术分为5个任务并记录住院医师、研究员和主治医师的主动控制时间(ACT)。结果显示住院医师在食管裂孔分离、胃底游离和胃底折叠任务中ACT百分比最高,研究员在所有5个任务中ACT均超过80%,且在纵隔分离和脚部修补任务中住院医师与研究员ACT存在显著差异。研究提示自动化任务级分割可客观量化受训者参与度。

要点

  • 住院医师在食管裂孔分离、胃底游离和胃底折叠任务中ACT百分比最高。
  • 研究员在所有5个任务中ACT均超过80%,胃底游离和食管裂孔分离最高。
  • 纵隔分离和脚部修补任务中住院医师与研究员ACT存在显著差异。

核心结论

  • 客观性能指标和自动病例分割可任务级量化受训者参与度,为培训进阶提供路线图。

研究设计与样本

  • 回顾性分析机器人辅助食管裂孔疝修补术的手术表现数据,病例经AI辅助标注和人工审核分为5个任务。

主要发现

  • 住院医师在食管裂孔分离(53%)、胃底游离(84%)和胃底折叠(57%)任务中ACT百分比最高。
  • 研究员在所有5个任务中ACT均超过80%,胃底游离(100%)和食管裂孔分离(88%)最高。
  • 纵隔分离和脚部修补任务中住院医师与研究员ACT存在显著差异。

临床怎么用

  • 可利用任务级ACT数据告知住院医师在其培训阶段预期参与的手术步骤。
  • 自动化评估对主治医师和住院医师工作流程影响小,可客观评估自主性。

局限性

  • 回顾性研究设计,任务分割错误病例被排除,可能引入选择偏倚。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

INTRODUCTION: Surgical autonomy for trainees has remained elusive to quantify. Proportion of active control time (ACT) of a trainee during a robotic case can be used as a broad measure of autonomy. However, this metric lacks in the granular detail of quantifying at what specific steps trainees were actively participating. We aim to quantify trainee involvement during robotic-assisted hiatal hernia repair at a task-specific level. METHODS: We performed a retrospective review of surgical performance data from robotic-assisted hiatal hernia repairs performed. These cases were segmented into 5 tasks by AI-assisted annotation with human review. The segmented tasks included: hiatal dissection, gastric fundus mobilization, mediastinal dissection, cruroplasty and fundoplication. Tasks were excluded if video segmentation of tasks was incorrect. During each task, ACT was recorded for resident, fellow and attending. Resident and fellow ACT per task was compared using the Mann-Whitney U test. RESULTS: Residents had the highest %ACT in the hiatal dissection (53%), gastric fundus mobilization (84%) and fundoplication (57%) tasks. Fellows had greater than 80% ACT in all 5 tasks, with the highest %ACT in the gastric fundus mobilization (100%) and hiatal dissection (88%). There was a significant difference between resident and fellow ACT during mediastinal dissection and cruroplasty. CONCLUSIONS: This study demonstrates how objective performance metrics and automated case segmentation can quantify trainee participation at a task-specific level. By utilizing data afforded by a robotic surgery platform, we are able to provide an objective and automated form of assessment with minimal impact on the workflow of attendings and residents. Our findings can serve to inform residents on what steps they can expect to be involved in during the procedure, appropriate to their PGY level. With this task-level data, we can provide a roadmap for trainee progression to achieve full surgical autonomy.

来源:Surg Endosc,查看原文。版权归原作者所有。

引用

  1. Gerull WD, Liebendorfer A, Awad MM, et al. Automated task-level autonomy assessment in robotic surgery. Surg Endosc. 2024. PMID: 39110222. doi:10.1007/s00464-024-11134-w

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。