Robotic emergency cholecystectomy tied to fewer bile duct injuries; AI models misjudge critical view of safety; AI speeds bleeding recognition

An instrument tray lined with a blue towel, holding a scalpel with an amber handle, toothed forceps and curved scissors laid side by side.

A Premier database study of 522,629 patients in the Journal of the American College of Surgeons linked robotic emergency cholecystectomy to fewer bile duct injuries than laparoscopy. A second study found vision-language models narrated cholecystectomy video fluently but judged the critical view of safety unreliably, and a preclinical trial reported faster bleeding recognition with AI assistance.

  • Research Oct 12, 2026
    In a propensity-matched Premier Healthcare Database study of emergency or urgent cholecystectomy from 2019 to 2024, robotic use rose from 4.2% to 29.7%. Bile duct injury through 30 days was 0.06% after robotic surgery versus 0.11% after laparoscopy (odds ratio 0.56, 95% CI 0.39 to 0.81), and conversion was 0.7% versus 2.1%, while operative time was longer (120.7 versus 109.4 minutes). The authors, one of them affiliated with Intuitive Surgical, said residual confounding remains possible.
  • AI Oct 13, 2026
    Researchers tested Gemini 2.5 Flash and a surgically trained model, Qwen3-VL-CoA, on 48 segments of 10 minutes from 13 Cholec80 videos, six of which showed the critical view of safety. With specific prompting, the untrained model had sensitivity of 0.33 and specificity of 0.95 for the critical view, and the surgically trained model had sensitivity of 0.83 and specificity of 0.19. The authors said fluent narration does not establish reliable safety assessment and that safety-critical judgments require separate task-specific validation.
  • AI Oct 13, 2026
    A model trained on 257 laparoscopic videos from 20 centers in China detected and tracked bleeding sites with an F1-score of 0.860. In a preclinical randomized trial, 77 surgeons were assigned to AI assistance or none; the primary end point, time from bleeding onset to the first successful click on the bleeding site, was 2.53 seconds with AI versus 3.23 seconds without (P < .001). The trial measured recognition on recorded video, not patient outcomes.
  • FDA Oct 14, 2026
    Gore said Oct. 14 that the Food and Drug Administration (FDA) approved an expanded indication for its Viabahn VBX balloon expandable endoprosthesis for use in fenestrated and branched endovascular aortic repair (FEVAR and BEVAR). The company said the approval gives U.S. vascular surgeons the first on-label bridging stent graft for both procedures. In three-year EMBRACE Registry results, vessel-level primary patency was 97.5% in 103 FEVAR patients and 94.0% in 165 BEVAR patients, according to the company.
    Source: Gore
  • Technology Oct 12, 2026
    A single-surgeon retrospective cohort used the da Vinci 5 platform's integrated force feedback instruments to record tension while closing the crura. Peak force was higher for hernias larger than 2.0 cm than for smaller ones (10.38 versus 8.78 N, P = .011). The authors said crural tensiometry is feasible within standard robotic workflow and that their exploratory stiffness index requires prospective validation against clinical outcomes before it can guide operative decisions. One author is a full-time Intuitive Surgical employee.

The weekly innovation roundup for general surgery, published every Sunday for the week before; every item links to its sources. Latest general surgery news · All specialties