Using AI during colonoscopy increases polyp detection rates and reduces costs

Using AI during colonoscopy increases polyp detection rates and reduces costs

A polyp is a small growth of tissue in the colon that can develop into colon cancer. Patients with polyps have a higher risk of developing cancer in the future.

• Der Einsatz künstlicher Intelligenz (KI) während der Koloskopie, einer endoskopischen Untersuchung mit einer Kamera, kann nützlich sein, um präkanzeröse Polypen zu erkennen, sagt Yuichi Mori, Arzt und außerordentlicher Professor der Gruppe für klinische Wirksamkeitsforschung an der Universität Oslo.

The AI ​​tool used in colonoscopy is based on the same technology as facial recognition on mobile phones. By marking the polyps in red or green boxes, the tool assists doctors during colonoscopy and helps them detect polyps that they would otherwise have missed.

“The technology is useful when the polyps are small or perhaps difficult to detect,” says Mori.

For more patients, a check-up colonoscopy is recommended every three years

A possible downside to the new technology is that doctors will use the tool to detect more small, harmless polyps, which then need to be removed. Even if the removal of small polyps does not cause any major side effects, one in thousands will experience bleeding or a minor rupture of the colon during removal.

Mori and co-researchers compared the results after colorectal cancer screenings with and without the use of AI. Around 5,800 patients were included in the study.

– The use of AI tools during colonoscopy led to a 20 percent increase in Europe in patients with polyps who were recommended screening every three years, says Mori. Normally, patients are recommended to have check-ups every 10 years.

The research was published in Impact of Artificial Intelligence on Colonoscopy Surveillance After Polyp Removal: A Pooled Analysis of Randomized Trials.

More screening can lead to overdiagnosis

As patients undergo more examinations, the use of AI during colonoscopy can lead to overdiagnosis, burden on patients and higher financial costs for healthcare services. Some patients also find the colonoscopy uncomfortable. Taking laxatives and drinking two to three liters of clear fluids before the exam may be difficult.

We don't know whether more colonoscopies lead to a lower risk of developing colon cancer over time, or whether it just increases the financial burden on hospitals. This requires longitudinal studies of clinical effectiveness and studies of the financial costs of using AI in colonoscopies.”

Yuichi Mori, Doctor and Associate Professor, Clinical Effectiveness Research Group, University of Oslo

Lower costs when using artificial intelligence in screening

In another new paper, Mori and his co-researchers have examined whether the use of AI in colorectal cancer screening actually leads to higher financial costs for health services. The study shows that the use of AI tools in colorectal cancer screening in the US reduced costs by $57 per person.

– The use of AI during screening can prevent approximately 7,200 cases of colorectal cancer and approximately 2,000 related deaths in the American population. This will save society $290 million per year in long-term costs. The study is based on modeling and we do not yet know whether the estimates are correct. To find out, we need large-scale clinical trials, says Mori.

The research was published in “Cost-effectiveness of Artificial Intelligence for Screening Colonoscopy: a Modeling Study.”

Mori is currently working on the EU-funded research project OperA, a portfolio of large-scale clinical trials on the long-term effects of using AI during screening.

• Ziel ist es, den Behandlungsstandard durch die Schaffung verlässlicher klinischer Beweise zu ändern und langfristig das Ergebnis für die Patienten zu verbessern, beispielsweise bei der Erkennung von Darmkrebs. Das OperA-Projekt werde die Antwort auf diese wichtigen Fragen finden, sagt Mori.

New colon cancer screening program in Norway

Norway recently introduced a national screening program to detect colorectal cancer. Starting in the fall, 55-year-old men and women will be offered screening using a colonoscopy or stool samples.

“I am pleased that Norway is now introducing a colorectal cancer screening program and I hope it will help reduce cancer,” says Mori.

“We don’t yet know whether AI can help make a difference in screening, but I hope it will be used when it can be useful,” he concludes.

Source:

University of Oslo, Faculty of Medicine

References:

• Mori, Y., et al. (2022) Einfluss künstlicher Intelligenz auf die Koloskopieüberwachung nach Polypenentfernung: Eine gepoolte Analyse randomisierter Studien. Klinische Gastroenterologie und Hepatologie. doi.org/10.1016/j.cgh.2022.08.022.
• Areia, M., et al. (2022) Kosteneffizienz künstlicher Intelligenz für die Screening-Koloskopie: eine Modellstudie. Die Lancet Digital Health. doi.org/10.1016/S2589-7500(22)00042-5.