AGA issues new recommendations on AI-assisted colonoscopy technology

The American Gastroenterological Association (AGA) released a new clinical guideline making no recommendation - for or against - the use of computer-aided detection systems (CADe) in colonoscopy. A rigorous review of evidence showed that artificial intelligence-assisted technology helps identify colorectal polyps. However, its impact on preventing colorectal cancer - the third most common cancer worldwide - remains unclear. 

Colonoscopy, performed more than 15 million times annually in the U.S., is an effective tool for detecting and preventing colorectal cancer. CADe systems have been shown to improve polyp detection rates, but whether that translates to reduced cancer cases is as yet unknown. 

"We are confident that using AI will lead to more polyps removed and more colonoscopies," said guideline author Benjamin Lebwohl, MD, AGAF, who said he encouraged his institution to adopt the technology. "We're less sure about the extent to which it will lead to less colon cancer. AI-assisted colonoscopy technology is promising and exciting. It's reasonable for practitioners to use the tech now, but we're not yet at a point where we can recommend universal adoption." 

With a growing number of studies evaluating the impact of AI-driven polyp detection, AGA is the first gastroenterological society in the U.S. to tackle an AI guideline for polyp detection. Data were evaluated using the rigorous GRADE process, and the final recommendation reflects significant input from GIs in community and academic practices. The guideline highlights key knowledge gaps that future studies need to address. 

Currently, CADe systems predominantly drive up the detection of low-risk polyps, which may result in more frequent and costly follow-up colonoscopies with uncertain benefits in preventing cancer, guideline authors said. Widespread adoption could also strain resources, limiting access for high-risk patients who need colonoscopies most. 

If AI is going to be impactful, it needs to be better than the human eye. Right now, AI is detecting easy-to-detect lesions. This is version 1.0. Before we can recommend everyone use AI, we need version 4.0, where it helps detect polyps that are truly difficult to find." 

Shahnaz Sultan, MD, MHSc, AGAF, guideline author

Next steps: Key knowledge gaps and future research 

AGA plans to update the guideline in one to two years as more data linking the use of CADe in colonoscopy to improve patient outcomes becomes available. Key areas for future research include: 

• Practitioner guidance: Clinicians should not feel obligated to use CADe but are encouraged to start as AI systems improve over time. 

• Quality over quantity: The focus should be on patient outcomes, such as post-colonoscopy colorectal cancer rates, rather than just polyp detection. 

• Rethinking surveillance: As CADe increases polyp detection, guidance on follow-up colonoscopy intervals should be reassessed. 

• Transparency in AI research: More publicly available data is needed to ensure AI models are rigorously compared and improved. 

Understanding colorectal cancer:

Colorectal cancer (also known as colon cancer) is cancer of the colon and/or rectum and occurs when a growth in the lining of the colon or rectum becomes cancerous. It is the third most common cancer worldwide and ranks as the third most common cause of cancer-related death in the United States. Most colorectal cancers come from precancerous polyps - adenomatous polyps or serrated polyps - that form over years (10 years on average) to become a cancer. A polyp is a mushroom-like or flat growth on the inside wall of the colon or rectum. Polyps grow slowly over many years. Not all colon polyps have the same risk of turning into colon cancer. Precancerous polyps could become cancerous; other types of polyps (hyperplastic, inflammatory) do not.