Surg Endosc. 2026 Jun 15. doi: 10.1007/s00464-026-12957-5. Online ahead of print.
ABSTRACT
BACKGROUND: Given the recent acquisition of the Fundamentals of Robotic Surgery (FRS) by the Society of Gastrointestinal and Endoscopic Surgeons (SAGES), there is a need to develop a proficiency-based training paradigm for the console-based tasks. The purpose of this study was to establish defensible, expert-derived training benchmarks for the FRS virtual reality (VR) tasks.
METHODS: A known-group standard setting framework was utilized. Five fellowship-trained minimally invasive surgeons (> 250 robotic cases) performed one warm-up and two recorded repetitions of each FRS VR task on the da Vinci SimNow platform. Times to completion and total scores were aggregated to determine the measures of central tendency for training benchmarks. To establish validity evidence, first-attempt performance of novices (fourth-year medical students enrolled in a simulation-based elective) was compared to the expert-derived mean time and median total score using one-sample tests.
RESULTS: Puzzle Piece Dissection was the most time-consuming task among experts (300 ± 61 s) and novices (584 ± 181). Knot Tying was the lowest-scoring task and demonstrated the greatest variability among experts (median score 54, IQR = [0-89]), whereas Ring Tower Transfer was the lowest-scoring for novices (median score 1, IQR = [0-49]). The differences between novice performance and expert measures of central tendency were statistically discernible (p < 0.05) with a large effect size for all tasks. Thus, training benchmarks were set at less than or equal to the expert-derived trimmed mean time and greater than or equal to the expert-derived median score.
CONCLUSION: Time- and score-based training benchmarks were established for the FRS VR tasks. The low and highly variable scores in Knot Tying were likely due to poor interaction fidelity, indicating that software modifications and/or alternative, non-VR exercises may be required for training. Further studies to evaluate the effectiveness of these training benchmarks are currently ongoing.
PMID:42298033 | DOI:10.1007/s00464-026-12957-5
