Related Experiment Videos
Applications of artificial intelligence along the perioperative pathway in metabolic and bariatric surgery
Karanbir Brar1, Inaam Chattha2, Samuel Gobraeil2
1Division of General Surgery, Department of Surgery, University of Toronto, Toronto, Ontario.
Background:
The use of artificial intelligence (AI) has rapidly increased in metabolic and bariatric surgery (MBS) in recent years, necessitating a comprehensive review characterizing the landscape of AI across the perioperative pathway of MBS care.
Objectives:
In this scoping review, we report on the applications of AI in the preoperative, intraoperative, and postoperative phases of care in MBS.
Setting:
Scoping review including articles published internationally.
Methods:
We systematically searched MEDLINE, Embase, Web of Science, and the Cochrane Database from inception until November 2024 for studies evaluating AI in any area of MBS. Studies were screened and extracted in duplicate, and a narrative synthesis of included studies was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews checklist.
Results:
We identified 58 studies for inclusion, with the majority of studies evaluating the applications of AI in the postoperative (35/58, 60.3%) phase, followed by the intraoperative (7/58, 12.1%) and preoperative (4/58, 6.9%) phases. A further 11/58 (18.9%) of studies evaluated large language models (LLMs) in MBS. Neural networks were the most frequently described algorithm (used in 26/47, 55.3% of studies), with LLM studies most frequently evaluating ChatGPT (10/11, 90.9%). Studies demonstrated significant promise in the ability of AI to accurately predict postoperative outcomes and support preoperative and intraoperative decision-making, and LLM studies showed the promise of AI in improving patient education and clinical decision support. However, the vast majority of studies were limited by minimal external validation and lack of direct prospective clinical evaluation.
Conclusions:
While AI shows significant promise in MBS, the existing literature is limited by minimal clinical evaluation and lack of external validation. Future prospective large-scale studies of AI across the perioperative pathway in MBS are required to demonstrate the utility of these algorithms in improving MBS care and patient outcomes.
