Artificial Intelligence-Aided Colonoscopy Does Not Increase Adenoma Detection Rate in Routine Clinical Practice

Area of Science:

• Gastroenterology
• Medical Artificial Intelligence
• Endoscopy

Background:

• Real-world performance data for artificial intelligence-aided colonoscopy (AIAC) is limited.
• AIAC systems aim to enhance polyp and adenoma detection during colonoscopies.
• Understanding AI's impact in clinical practice is crucial for adoption.

Purpose of the Study:

• To evaluate the impact of introducing AIAC (GI Genius, Medtronic) on adenoma and polyp detection rates (ADR/PDR) in a large-volume center.
• To compare colonoscopy performance metrics before and after AIAC implementation.
• To assess the real-world effectiveness of AIAC in a clinical setting.

Main Methods:

• A 6-month observational study comparing ADR/PDR before (pre-AIAC) and after AIAC introduction.
• AIAC was implemented across all endoscopy suites in a high-volume medical center.
• Adenoma and polyp detection rates, and procedure times were the primary outcome measures.

Main Results:

• Adenoma detection rate (ADR) was lower in the AIAC group (30.3%) compared to the pre-AIAC group (35.2%) (P < 0.001).
• Polyp detection rate (PDR) was also lower with AIAC (36.5%) versus pre-AIAC (40.9%) (P = 0.004).
• Procedure time was significantly shorter in the AIAC group.

Conclusions:

• The introduction of AIAC did not lead to improved adenoma or polyp detection in this large-center cohort.
• Shorter procedure times were observed with AIAC, but this did not correlate with enhanced diagnostic yield.
• The findings raise important questions regarding AI-human interaction and the practical implementation of AIAC in clinical practice.