Feed

Superficial gastrointestinal (GI) neoplasms pose a significant global health burden, with endoscopic submucosal dissection (ESD) as the gold standard for curative resection. However, conventional ESD (C-ESD) is technically challenging, with prolonged procedure times and higher complication risks. Robotic-assisted ESD (R-ESD) offers enhanced precision and safety, but evidence is limited. To evaluate the efficacy and safety of R-ESD using a Bayesian single-arm Meta-analysis. Systematic review and Bayesian single-arm meta-analysis. Following PRISMA guidelines and PROSPERO registration (CRD420251273787), we conducted a systematic review and Bayesian single-arm meta-analysis of randomized/nonrandomized comparative and single-arm studies on R-ESD for superficial GI neoplasms. Searches spanned MEDLINE, Embase, Scopus, and others through December 2025. Primary outcomes: en-bloc and R0 resection rates. Secondary outcomes: procedure time, perforation, bleeding. Bayesian random-effects models used non-informative priors; analyses in R with the brms package. Risk of bias was assessed via RoB 2 and NOS; certainty via GRADE. Five studies (183 lesions) were included. Frequentist pooled en-bloc resection rate: 98.2% (95% CI: 0.916-0.996, I2 = 0%); R0: 96.8% (95% CI: 0.759-0.997, I2 = 47.8%). Bayesian medians: en-bloc 93.4% (95% CrI: 0.856-0.971); R0 90.0% (95% CrI: 0.798-0.957). 77.6 min (95% CI: 45.0-110.3, I2 = 99%); perforation: 1.2% (95% CI: 0.002-0.077, I2 = 0%); bleeding: 10.6% (95% CI: 0.019-0.416, I2 = 30.8%). Evidence certainty:low. R-ESD demonstrates promising efficacy and safety, but current evidence is limited to small, mostly non-comparative studies. Larger comparative trials are required before definitive conclusions can be made.