Translating artificial intelligence into clinical practice for gastrointestinal endoscopy: current applications and...

Area of Science:

• Gastroenterology and Endoscopy
• Medical Artificial Intelligence
• Diagnostic Technologies

Background:

• Artificial intelligence (AI) is revolutionizing gastrointestinal (GI) endoscopy.
• AI addresses key challenges in lesion detection, diagnosis, and clinical decision-making.

Purpose of the Study:

• To review the current applications and impact of AI in various GI endoscopic procedures.
• To assess the performance, limitations, and future directions of AI in GI endoscopy.

Main Methods:

• Review of AI applications in upper GI endoscopy, lower GI endoscopy, inflammatory bowel disease, and pancreatobiliary endoscopy.
• Analysis of AI's role in detection, diagnosis, scoring, and interpretation across different endoscopic modalities.

Main Results:

• AI shows high accuracy in upper GI endoscopy for H. pylori and premalignant/malignant lesions.
• Computer-aided detection/diagnosis improves adenoma detection and management strategies in lower GI endoscopy.
• AI standardizes scoring in IBD and shows strong performance in pancreatobiliary applications, but real-world benefits are modest.
• Challenges include overdetection, variable performance, explainability, workflow integration, and cost-effectiveness.

Conclusions:

• AI demonstrates expert-level performance in multiple GI endoscopic domains.
• Further multicenter validation and addressing implementation challenges are crucial for AI adoption.
• AI has the potential to become integral to routine endoscopic practice.