BACKGROUND AND AIMS: Visual gaze pattern (VGP) analysis quantifies endoscopists' specific eye movements. VGP during colonoscopy may be associated with polyp detection. However, the optimal VGP to maximize detection performance remains unclear. This study evaluated the optimal endoscopic VGP that enabled the highest colorectal adenoma detection rate. METHODS: This randomized controlled trial was conducted between July and December 2023. We developed an eye-tracking and feedback (ETF) system that instructed endoscopists to correct their gaze toward the periphery of an endoscope screen with an audible alert. Patients who underwent colonoscopy were randomly assigned to 4 groups: 3 intervention groups, in which the endoscopist's gaze was instructed to a different level of the peripheral screen area using the ETF system (the periphery of 4 × 4, 5 × 5, and 6 × 6 divisions of the screen), and a control group in which the endoscopist did not receive instructions. The primary outcome was the number of adenomas detected per colonoscopy (APC). RESULTS: In total, 189 patients were enrolled. The APC and adenoma detection rate were significantly higher in the 6 × 6 group than in the control group (1.82 ± 2.41 vs 0.59 ± 1.17, P = .002
68.9% vs 30.8%, P = .002). The APC and the number of screen divisions were positively correlated (R = 0.985, P = .0152). The rate at which the endoscopist gazed at the periphery of the screen was positively correlated with the number of divisions (R = 0.964, P = .0363). CONCLUSIONS: Colorectal adenoma detection was improved by correcting the endoscopist's gaze to the periphery of the screen, especially by dividing the screen into 6 × 6 segments.