A cloud-based artificial intelligence system, named Caddie, has shown significant improvements in the detection of adenomas, including those that are hard to detect in the proximal colon. This was especially evident when used by non-expert endoscopists in the UK. The findings were presented at UEG week. Rawen Kader, MRCP, PhD, a clinical research fellow at University College London, emphasized that while many previous trials in this field were conducted by expert endoscopists, the main advantage of AI in polyp detection was hoped to benefit non-expert endoscopists.
Key findings from the study include:
- The adenoma detection rate was 8% higher with Caddie compared to the standard of care.
- Proximal ADR increased from 15% to 23% with Caddie.
- The study involved 739 adults scheduled for surveillance or symptomatic colonoscopy, with 614 included in the analysis. Procedures were performed by 26 endoscopists at nine academic and community hospitals in the UK.
- The adenoma detection rate for non-expert endoscopists using Caddie increased from 25% to 33%.
- The odds of detecting adenomas were 50% higher with Caddie compared to the standard of care.
- The significant increase in detection with Caddie was especially evident in harder-to-detect polyps, such as flat non-polypoid sessile polyps found in the proximal colon.
Rawen Kader also highlighted that each increase in ADR correlates with a reduced risk of a patient developing post-colonoscopy colorectal cancer (PCCRC). The risk of PCCRC is higher in non-expert endoscopists, and AI could be crucial in reducing this variability in ADR and the prevalence of PCCRC.