MyIntuitive遥测识别机器人胰十二指肠切除术的熟练度阈值
本研究分析单术者连续100例机器人胰十二指肠切除术(RPD),以MyIntuitive术中遥测的空闲时间为主要终点,结合风险调整累积和(RA-CUSUM)与样条逻辑回归评估学习曲线。结果显示空闲时间、控制台时间等指标在约60例后显著改善,转换率从23.3%降至7.5%,提示遥测可敏感识别熟练度阈值。
要点
- MyIntuitive遥测空闲时间可敏感追踪RPD学习曲线。
- RA-CUSUM显示熟练度拐点约在50-60例。
- 转换率从23.3%降至7.5%(P=0.033)。
核心结论
- 基于遥测的主要终点(MyIntuitive空闲时间)可敏感追踪RPD学习曲线,结合风险调整CUSUM识别熟练度阈值约60例。
研究设计与样本
- 单术者连续100例机器人胰十二指肠切除术尝试(2022年1月至2025年8月),主要终点为完成机器人病例的全病例空闲时间,次要终点包括控制台时间、活动时间、机械臂活动及器械交换,并以意向性治疗(ITT)分析转换概率。
主要发现
- 初始阶段(1-60例)与熟练阶段(>60例)相比,空闲时间从42分钟降至32分钟(P=0.001),控制台时间从470分钟降至395分钟(P<0.001),活动时间从428分钟降至365分钟(P<0.001),第4臂活动时间从40分钟降至17分钟(P=0.027)。
- 空闲时间的RA-CUSUM显示熟练度拐点接近50-60例;支持性切除阶段遥测显示完成胰周间隙解剖时间从323分钟缩短至215分钟(P=0.002)。
- ITT分析中总体转换率为17%,第60例后从23.3%降至7.5%(P=0.033),无急诊转换。
- 围手术期结局包括术后胰瘘29%、胰腺切除后出血8%、Clavien-Dindo III级及以上22%、再手术9%、90天再入院9%、90天死亡率2%,中位住院时间16天。
临床怎么用
- 可利用MyIntuitive遥测空闲时间等过程指标客观追踪机器人胰十二指肠切除术的学习曲线,指导培训与项目成熟度评估。
局限性
- 单术者、单中心研究,样本量有限,结果外推需谨慎。
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
BACKGROUND: Quantifying learning in robotic pancreaticoduodenectomy (RPD) can be challenging because case selection and conversions complicate the interpretation of full-procedure metrics. MyIntuitive intraoperative telemetry provides objective, process-level signals that may help to capture proficiency gains. METHODS: We analysed 100 consecutive RPD attempts by a single surgeon (January 2022-August 2025). The primary outcome was Full-Case idle time from MyIntuitive telemetry among completed robotic cases. Key secondary telemetry outcomes included console time, active time, arm-specific activity, and instrument exchanges. As secondary corroboration on an intention-to-treat (ITT) basis, the probability of robotic completion (no conversion) was modelled across case order with risk-adjusted (RA)-CUSUM and spline logistic regression; conversions were characterised by timing and reason on video review. RESULTS: Telemetry demonstrated marked efficiency gains across the curve. Idle time fell from 42 to 32 min (P = 0.001), console time from 470 to 395 min (P < 0.001), and active time from 428 to 365 min (P < 0.001) between the initial (cases 1-60) and proficiency (> 60) phases; Arm-4 active time decreased from 40 to 17 min (P = 0.027). The RA-CUSUM of idle time showed a proficiency inflection near cases 50-60. Supportive resection-phase telemetry showed shorter time to complete mesopancreas dissection (from 323 to 215 min, P = 0.002). On ITT analysis, conversion occurred in 17% overall, declining from 23.3 to 7.5% after case 60 (P = 0.033); no emergency conversions occurred. Perioperative outcomes included POPF 29%, postpancreatectomy haemorrhage 8%, ≥ Clavien-Dindo III 22%, reoperation 9%, 90-day readmission 9%, 90-day mortality 2%, and median length of stay 16 days. CONCLUSION: A telemetry-based primary endpoint (MyIntuitive idle time) sensitively tracks the RPD learning curve and, with risk-adjusted CUSUM, identifies a proficiency threshold around ~ 60 cases. Secondary ITT completion trends and conversion mapping corroborate the telemetry signal. This telemetry-first framework provides objective, phase-specific insight to guide coaching and programme maturation in complex robotic pancreatic surgery.
引用
- González-Abós C, Reyes LG, Lorenzo C, et al. MyIntuitive telemetry identifies proficiency thresholds in robotic pancreaticoduodenectomy. Surg Endosc. 2026. PMID: 41642310. doi:10.1007/s00464-026-12591-1
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。