Getting a colonoscopy? Ask your doctor about using an AI copilot

Colon cancer can often be prevented through timely screening. One way to get screened is through a procedure called colonoscopy, in which a doctor examines your colon with a camera to look for precancerous polyps. Unfortunately, some doctors are not as thorough as they should be and miss growths that they should remove. It might be time to ask if your doctor is ready to get help from AI.

To understand why AI could help, imagine having to choose between two doctors to perform your colonoscopy. After the procedure is explained, the first reassures you that there is no need to worry, as precancerous lesions are rare: “I only discover polyps 10% of the time in patients like you, and then remove them before they become cancerous .” The 10% figure is intended to reassure you about how unlikely it is that you will develop cancer or pre-cancerous lesions in your colon.

The second doctor also tries to reassure you, but with a notable difference: this doctor explains that he discovers polyps in patients like you ‘30% of the time’.

All else being equal, which doctor would you choose?

At first glance, it may seem like the first doctor is the best, given the much lower risk of pre-cancer in their patients. However, I advise you to choose the second one. You see, some doctors are too quick to scan the colon for polyps. Perhaps they are not particularly thorough by nature. Or perhaps they feel pressure to maximize their revenue by spending less time with their patients. Either way, these doctors are more likely to miss polyps, and thus their patients are more likely to develop cancer later.

Here’s one image to illustrate the relationship between the detection of polyps (i.e. adenoma) and the likelihood of later being diagnosed with colon cancer (what the cool kids call PCCRC: post-colonoscopy colorectal cancer). As a doctor’s adenoma detection rate, or ADR, increases, the likelihood that their patients will later be diagnosed with colon cancer decreases.

Association between adenoma detection rate by physician and risk of postcolonoscopy colorectal cancer

In other words, doctors who find fewer polyps miss growths they should remove.

What should we do about these hasty colonoscopists? For policymakers, this finding suggests that quality measures for physicians who perform colonoscopies should identify those who often fail to detect polyps in their patients. In addition, we need to publish polyp detection rates so that patients can make informed decisions about which doctors to seek out to perform this life-saving test.

But there is something else patients can do. Ask your doctors if they use AI assistance during the procedure. Studies in Italy and the US. show a significant increase in polyp detection when doctors use AI assistance. Essentially, they have a computer scan the camera footage, live and highlight suspicious areas so they can take a closer look.

I expect that the use of AI-based computer copilots will soon become routine. In the meantime, consider asking your doctor if they use such a program. Even the best doctors make mistakes. As any airline will tell you, sometimes it pays to have copilots.