Evidence-Based Prevention of De Novo GERD after Bariatric Surgery: Comparing Human and AI Inference

With the global rise in obesity rates, metabolic and bariatric surgery (MBS) has become the primary treatment modality for managing obesity and its associated complications. Gastroesophageal reflux disease (GERD), a common complication of obesity primarily caused by increased intra-abdominal pressure and impaired lower esophageal sphincter function, often shows improvement following MBS. However, paradoxically, MBS can also lead to the development of de novo GERD in some patients [ 1 ]. Although de novo GERD can be an inherent consequence of MBS procedures, particularly those that are naturally refluxogenic [ 2 , 3 ], the risk is not solely procedure-dependent. Technical refinements during surgery can play a crucial role in preventing or reducing the incidence of de novo GERD. Furthermore, some patients may be inherently more susceptible to developing GERD postoperatively, highlighting the importance of careful patient selection [ 4 , 5 ]. Standardization of preoperative evaluation, diagnostic definitions, and risk assessment protocols is also essential to improve preventive strategies [ 6 , 7 ]. Despite its clinical relevance, there is currently no established guideline or consensus on how best to prevent de novo GERD following MBS. With the advent of artificial intelligence (AI), particularly large language models (LLMs), clinicians now have powerful tools to access, interpret, and apply vast amounts of clinical data [ 8 ]. These technologies can support surgeons in selecting suitable candidates, selecting the optimal surgical procedure, and refining technical details to minimize the risk of de novo GERD. This study aims to provide an evidence-based framework for the diagnosis and prevention of de novo GERD after MBS, while also comparing human clinical reasoning with AI-driven inference in this context.