A surgical workflow ontology for endometriosis surgery: systematic derivation and international modified Delphi consensus

Background. Endometriosis surgery has no agreed vocabulary of operative steps. Before an operation can be recognised automatically in video, it must first be described; and surgical data science has repeatedly identified the absence of an accepted annotation standard as the field's rate-limiting problem. Objective. To derive, from the published literature, a candidate set of operative-step items for deep endometriosis surgery and minimally invasive hysterectomy, and to submit that set to an international modified Delphi consensus, producing a surgical workflow ontology with operational start and end triggers. Design. A two-phase methodological study. Phase 1: a systematic literature search, conducted according to PRISMA 2020, to derive the candidate item set. Phase 2: a modified Delphi with an international panel of 18 seats, conducted and reported according to ACCORD, with agreement thresholds fixed a priori. Primary outcome. The number and proportion of voting-instrument questions (n = 224) reaching strong agreement — at least 80% of responses in the 7–9 band — at the end of the planned rounds. As a derived secondary outcome, the number and proportion of ontology labels (n = 67) whose associated questions reached strong agreement, using the aggregation rule declared in section 5.4.4. Secondary outcomes. Proportion of items reformulated between rounds; items without consensus, reported as equipoise; global panel agreement (Cronbach's α and Kendall's W); and the correspondence map between the ontology, the AAGL 2021 classification and #Enzian.