Primary anastomosis or Hartmann’s procedure in complicated diverticulitis: the AIDE/OBS multicentre feasibility protocol for clinical and visual data in AI research

The choice between primary colorectal anastomosis and Hartmann’s procedure in non-elective minimally invasive surgery for complicated acute diverticulitis remains variable and depends mostly on surgeon’s intraoperative judgment. Standardized clinical and visual data may help characterize the findings associated with this decision and support future artificial intelligence research in emergency surgery. AIDE/OBS is an international, multicenter, observational study including retrospective and prospective cases of adults undergoing non-elective surgery for complicated acute diverticulitis in whom a laparoscopic or robotic approach is attempted. Eligible operations include Hartmann’s procedure and sigmoidectomy or left colectomy with primary anastomosis, with or without diversion. Participating centers will collect pseudonymized clinical, radiological, operative, surgeon-reported, and postoperative data and submit de-identified intraoperative videos and/or sets of still images labelled according to predefined operative phases. Videos and still-image sets will be assessed separately. The primary outcome is the feasibility of assembling linked clinical and intraoperative visual datasets. This will be assessed through clinical data completeness and the yield, analyzability, and operative-phase coverage of videos and still-image sets, reported separately. Secondary outcomes include factors associated with the observed operative strategy, stoma creation at the index operation, and anastomotic leak, morbidity graded by the Clavien–Dindo classification, unplanned reoperation, and all-cause mortality within 30 days. AIDE/OBS will assess the feasibility of collecting linked clinical data, videos, and still-image sets in this emergency surgical setting. The resulting phase-annotated datasets may inform exploratory image analysis and subsequent development of artificial intelligence tools; this study will not validate a clinical decision-support tool.