Key research questions for implementation of artificial intelligence in capsule endoscopy

Area of Science:

• Medical Artificial Intelligence
• Gastrointestinal Endoscopy
• Capsule Endoscopy

Background:

• Artificial intelligence (AI) is increasingly integrated into medical practice, with gastrointestinal endoscopy being a leading field.
• Challenges in adopting AI for endoscopy are currently being addressed in consensus documents.

Purpose of the Study:

• To identify future research challenges and areas for AI implementation in capsule endoscopy (CE).

Main Methods:

• A modified three-round Delphi consensus online survey was conducted with CE and AI experts.
• Experts mapped research statements and challenges, ranked them, and rescored top statements.

Main Results:

• Twenty-one experts participated, generating 48 statements across seven themes.
• The highest-ranked challenges included AI for small bowel pathology identification/grading (9.15), AI-human expert correlation (9.05), and real-life feasibility (9.0).

Conclusions:

• The study provides a roadmap for future research to fully integrate AI into CE reading.
• Key areas for AI development in CE include accurate pathology detection, seamless human-AI collaboration, and practical clinical application.