Statement offers pragmatic foundation on use of AI in GI endoscopy
An international consensus statement from the World Endoscopy Organization offered 10 recommendations in three domains: data governance, privacy, and algorithmic transparency; artificial intelligence (AI) algorithms and medical liability; and equity and bias.
The World Endoscopy Organization recently issued an international consensus statement addressing legal, ethical, and equity considerations for use of artificial intelligence (AI) in GI endoscopy. Fourteen experts from 11 countries developed the guidance, which was grouped into the domains of data governance, privacy, and algorithmic transparency; AI algorithms and medical liability; and equity and bias. The consensus statement was published Dec. 9 by Annals of Internal Medicine.
Regarding data governance, privacy, and algorithmic transparency, the panel stated that AI systems should comply with local data protection regulations, ensure patient privacy through anonymization, and maintain transparency in algorithm updates and performance reporting. The panel also said that hospitals and gastroenterology practices should establish clear policies on data ownership and usage rights in the endoscopy suite, ensuring that patients are aware of how their data will be used, particularly when third-party commercial AI vendors are involved.
In the domain of medical liability, the panel said that physicians must use AI tools within the bounds of accepted clinical practice and in accordance with manufacturer guidance. Professional societies should provide clear liability guidance for AI-assisted diagnosis and automated reporting, the panel noted. “Before the widespread adoption of AI-driven automated report generation and novel AI-enabled performance indicators (for example, “percentage of mucosa evaluated” or “effective inspection time”) in GI endoscopy, it is essential to evaluate their accuracy, understand their clinical relevance, and clarify associated medicolegal implications,” they wrote.
Regarding equity and bias, the panel called for AI algorithms to be validated on diverse data sets, with transparent reporting of demographic characteristics to allow stakeholders to assess the technology's generalizability and equity. More research is needed to evaluate whether AI use in GI endoscopy risks exacerbating existing health care disparities and to guide effective bias mitigation where needed, the panel noted.
“The trajectory of AI innovation and adoption in GI endoscopy will depend heavily on evolving regulatory, legal, and health system frameworks,” the panel concluded. “These consensus statements are intended as a pragmatic foundation to support thoughtful, equitable, and evidence-based AI adoption in GI endoscopic practice worldwide.”