Bowel cancer usually develops over years from polyps, in particular from so-called adenomas. If they are detected and removed during a colonoscopy, cancer can be prevented. This is exactly where one of the best-studied AI applications in medicine comes in.
How AI helps during colonoscopy
In computer-aided detection (CADe), software analyses the colonoscopy video image in real time. If it detects a suspicious structure, a coloured box appears on the monitor. The doctor then decides whether it is a polyp and whether it should be removed.
An important quality measure is the adenoma detection rate (ADR): the proportion of examinations in which at least one adenoma is found. Studies show that the higher it is, the less often the people examined later develop bowel cancer. Numerous randomised trials show that CADe increases the ADR. A meta-analysis by Hassan and colleagues (2021) found an average ADR of 36.6 per cent with AI compared with 25.2 per cent without.
The downside: loss of skill
In August 2025, The Lancet Gastroenterology & Hepatology published an observational study from four endoscopy centres in Poland. The researchers compared colonoscopies performed without AI before and after the technology was introduced. The result: after the doctors had worked regularly with AI for a few months, their ADR in examinations without AI fell from 28.4 to 22.4 per cent.
The authors speak of possible “deskilling”: those who become used to the software pointing out abnormalities may search less attentively without it. This is an observational study that does not prove a definite cause-and-effect relationship. Nevertheless, it is one of the first studies to put numbers on this long-suspected risk – with implications far beyond colonoscopy.
Further open questions
- Small polyps: some of the additionally detected adenomas are very small. Whether removing them noticeably reduces the risk of cancer has not been conclusively established.
- Experienced endoscopists: those who already have a very high ADR may benefit less.
- Long-term benefit: long-term studies are needed to show whether CADe reduces the number of bowel cancer cases and deaths.
What this means for patients
Colonoscopy remains one of the most effective cancer screening examinations – with or without AI. In Germany, people with statutory health insurance are entitled to a screening colonoscopy from the age of 50, which can be repeated once after ten years at the earliest (if the first examination takes place at 65 or older, one examination is covered). Alternatively, there are stool tests for hidden blood. Good bowel preparation and experienced endoscopists are crucial for quality. If you wish, you can ask during the preliminary consultation whether an AI system will be used.
Conclusion
Colonoscopy exemplifies both sides of medical AI: measurably better detection, but also new risks when people rely too heavily on the technology. Professional societies are therefore discussing how training and continuing education can be designed so that clinicians’ own skills are maintained.
More on this topic: Risks of AI in medicine.
Note: This article is for general information only and does not replace medical advice. As of: October 2026.
Video on this article
Watch on YouTube · Subscribe to the channel
Sources
- HMP Global: AI Exposure May Reduce Polyp Detection Skills of Endoscopists (Lancet Gastroenterology & Hepatology, 2025)
- The Lancet Gastroenterology & Hepatology 2025: Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study
- Hassan C et al.: Performance of artificial intelligence in colonoscopy for adenoma and polyp detection: a systematic review and meta-analysis. Gastrointestinal Endoscopy 2021
- Leitlinienprogramm Onkologie: S3-Leitlinie Kolorektales Karzinom
- Gemeinsamer Bundesausschuss (G-BA): Richtlinie für organisierte Krebsfrüherkennungsprogramme
Leave a Reply