Beyond surgical showcase: workflow continuity and exploratory AI-oriented video interpretability of robotic low
1Department of Surgical Oncology, Ankara Bilkent City Hospital, Ankara, Türkiye.
Background:
Publicly available robotic surgical videos are increasingly used as supplementary educational resources in colorectal surgery. However, beyond general educational quality, the extent to which robotic low anterior resection (LAR) videos preserve workflow continuity, critical anatomical visualization, and expert-rated AI-oriented video interpretability remains unclear. This study aimed to evaluate publicly available robotic LAR videos on YouTube for workflow completeness, educational continuity, and exploratory AI-oriented video interpretability.
Methods:
A structured cross-sectional video-based analysis of publicly available robotic LAR videos on YouTube was performed using predefined robotic rectal surgery-related search terms. Workflow domains including setup phases, vascular dissection, pelvic dissection, total mesorectal excision (TME), critical anatomical visualization, reconstruction phases, and safety-oriented procedural steps were systematically evaluated. Workflow completeness and AI-oriented video interpretability were assessed using structured expert-rated frameworks, whereas educational value was summarized using an author-derived exploratory composite score. Associations between video characteristics and composite scores were analyzed.
Results:
Thirty-nine robotic LAR videos met the eligibility criteria. Median video duration was 647 s. Total mesorectal excision (TME) dissection, pelvic dissection, distal rectal transection, and inferior mesenteric artery/inferior mesenteric vein (IMA/IMV) dissection were among the most consistently demonstrated operative phases. In contrast, patient positioning, trocar placement, robotic docking, specimen extraction, and leak testing demonstrated substantially lower representation rates. Dunn's post-hoc comparisons with Bonferroni correction demonstrated higher workflow completeness, educational value, and AI-oriented interpretability scores in structured workflow videos than in highlight/showcase videos. Workflow analysis demonstrated marked heterogeneity across operative domains, with pelvic dissection-related phases showing the highest visibility rates. Viewer engagement metrics were not significantly associated with workflow completeness or AI-oriented interpretability.
Conclusions:
Publicly available robotic low anterior resection videos on YouTube predominantly emphasize visually prominent pelvic dissection phases while inconsistently preserving complete operative workflow continuity. Many robotic LAR videos may function more as selective technical demonstrations than as standardized workflow-oriented educational resources. These findings should be interpreted as video-level workflow and interpretability observations, not as evidence of effects on trainee performance, operative quality, patient safety, or actual AI model performance. Future robotic surgical video platforms may benefit from workflow-oriented reporting standards emphasizing procedural continuity, anatomical interpretability, reconstruction, and safety-oriented workflow representation.
