Man Plus Machine: A Randomized Control Trial of Artificial Intelligence Including the Impact of Adjunctive Polyp...

Area of Science:

• Gastroenterology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Computer-aided polyp detection (CADe) tools using artificial intelligence (AI) show promise for enhancing polyp detection during colonoscopy.
• The impact of CADe when used alongside other advanced techniques like increased withdrawal time, retroflexion, dynamic position change, and narrow-band imaging (NBI) requires further investigation.

Purpose of the Study:

• To evaluate the effectiveness of an AI-powered CADe module (ENDO-AID) compared to conventional colonoscopy in improving adenoma detection rate (ADR).
• To assess the combined impact of CADe with traditional polyp detection enhancement techniques.

Main Methods:

• A single-center prospective randomized controlled trial was conducted involving 776 patients.
• Endoscopists (n=26) were randomized to use either the ENDO-AID AI module or conventional colonoscopy.
• Data on the use of additional techniques (e.g., NBI, position change) were recorded.

Main Results:

• While univariate analysis showed a nonsignificant trend towards higher ADR with CADe (63.4% vs. 57.3%), multivariable analysis revealed CADe was independently associated with increased ADR (OR, 1.38; P=0.042).
• CADe demonstrated particular effectiveness in the screening cohort (ADR 79% vs. 68%).
• Use of NBI (OR, 2.00) and increased withdrawal time (OR, 1.11) were also independently associated with higher ADR.

Conclusions:

• AI-assisted CADe significantly increased adenoma detection rates, particularly in a cohort of high-performing endoscopists.
• The benefits of CADe were augmented by traditional techniques, highlighting the importance of integrating both approaches to maximize ADR.