ARISING HEALTHALIVE. AWAKE. AWARE.Read the magazines →

← November 19, 2026 edited reader edition

THE ARISING · Longevity & Healthspan

AI helped less-experienced colonoscopists detect more adenomas

Five randomized trials showed no clear gain in finding advanced adenomas, and did not establish fewer cases of cancer.

In five randomized trials, trainees and less-experienced doctors using computer assistance found adenomas more often than those using standard colonoscopy. No clear increase was shown for advanced adenomas.

The study is a systematic review and meta-analysis, published in 2026 in Gastroenterology Research. The authors combined five randomized controlled trials covering 3,158 patients. Each trial compared colonoscopy with computer-aided detection (CADe) against standard colonoscopy, performed by trainees or less-experienced endoscopists. CADe software flags suspicious areas on the screen as the doctor examines the colon.

The main measure was the adenoma detection rate. An adenoma is a type of polyp that can turn into colon cancer, so finding and removing them is a central goal of the exam. With CADe, the rate was higher by a risk ratio of 1.17 (95% confidence interval 1.07 to 1.29). That works out to an absolute gain of 6.9 percentage points. That works out to about 15 examinations for one additional patient with an adenoma detected, on average. It does not mean one cancer prevented.

The gain was larger in the right side of the colon. There, the risk ratio was 1.31 (95% confidence interval 1.14 to 1.50), an absolute gain of 8.2 percentage points. The overall polyp detection rate was also higher, with a risk ratio of 1.16.

Advanced adenomas are the larger or more worrying polyps. Here the picture was different. The detection rate did not differ between the CADe and standard groups (risk ratio 0.97, 95% confidence interval 0.73 to 1.29).

The overall result held up when the authors removed each trial in turn. It also held when they dropped the one trial of independently practicing, non-expert operators and looked only at doctors in formal training (risk ratio 1.16, 95% confidence interval 1.01 to 1.33).

The authors rated their confidence in the evidence as moderate for overall adenoma detection. It was low for right-colon and overall polyp detection, and very low for advanced adenomas. Their prediction intervals, which estimate the range of effects in future settings, crossed the point of no effect for the main results.

The additional findings in these trials were not shown to translate to fewer cancers or improved outcomes. The trials also say little about what happens when operators later perform the procedure without the software.

The authors conclude that the findings support further prospective evaluation of CADe as an adjunct to colonoscopy training. They say the results do not establish that building it into fellowship curricula produces more competent independent endoscopists.

What to watch next: larger trials in trainees and less-experienced operators, with a focus on whether CADe raises detection of advanced adenomas and right-colon polyps.

Sources

THE ARISING NEWSLETTER

Science moves forward.
Come with it.

Research, clear explanations, and a little room for wonder.
Read and follow The Arising on Substack.

Find The Arising on Substack ↗