Establishing key research questions for the implementation of artificial intelligence in colonoscopy: a modified...

Area of Science:

• Gastroenterology
• Medical Informatics
• Artificial Intelligence

Background:

• Artificial intelligence (AI) research for colonoscopy shows rapid progress.
• Widespread clinical implementation of AI in colonoscopy is not yet a reality.
• Identifying key research priorities is crucial for accelerating AI adoption in endoscopy.

Purpose of the Study:

• To identify the top implementation research priorities for artificial intelligence (AI) in colonoscopy.
• To establish a framework for future research and stakeholder collaboration.

Main Methods:

• A modified Delphi approach was employed for research priority setting.
• Fifteen international experts (endoscopists, computer scientists/engineers) from nine countries participated.
• A 9-month online survey process involved generating and scoring research questions.

Main Results:

• The top 10 research questions were categorized into five themes: clinical trial design/end points, technological developments, clinical adoption/integration, data access/annotation, and regulatory approval.
• Key priorities include optimizing trial designs, improving lesion detection and reducing false positives, integrating AI into workflows, enhancing data annotation, and streamlining regulatory processes.
• Specific focus areas include evaluating AI's impact on polyp detection, characterization, and interval cancer rates.

Conclusions:

• This study represents the first international research priority-setting exercise for AI in colonoscopy.
• The findings provide a framework to guide future research and stakeholder engagement.
• Accelerating the clinical implementation of AI in endoscopy is the ultimate goal.