Optimal bowel preparation is crucial for successful colonoscopy, yet inadequate preparation remains a significant issue. This study introduces a smartphone app utilizing an AI-based prediction system trained on fecal photographs to assess bowel preparation quality. A prospective, single-masked, multicenter randomized clinical trial compared the app group to a control group. The AI system achieved 95.15% accuracy and 0.98 AUC. The AI-driven app group showed significantly higher rates of adequate bowel preparation (88.54% vs. 65.59%), better compliance with instructions, and increased additional purgative intake. The app significantly improved bowel preparation quality and patient compliance.
