AI promises breakthrough in bowel cancer detection

Earlier diagnosis of bowel cancer is the promise of using artificial intelligence to analyse people's gut to predict whether they are at risk of the disease.

Researchers would train the AI on stool samples discarded after being tested as part of New Zealand's bowel screening programme.

The screening tests, called a Fecal Immunochemical Test, or FIT, look for traces of blood in a person's stool, which is a risk factor for bowel cancer. The test is designed to detect tiny amounts of blood not visible to the naked eye.

Research fellow at University of Auckland's Liggins Institute Dr Theo Portlock said once the test was done, AI could be trained on the remaining stool sample with the hope of identifying a person's risk factor with up to 90 percent accuracy, helping earlier diagnosis.

"This FIT test is used pretty much worldwide but people are gradually starting to see the importance of AI in understanding these very complicated relationship between different pieces of data and how they relate to bowel cancer risk," Portlock said.

"AI is very good at doing that understanding, something that we wouldn't be able to do with a pen and paper or on your computer traditionally.

"So it's the first time in recorded human history where this has been possible based on the algorithms that are available but also on the computer power that we have these days. So it's an incredibly exciting time to be in this research."

The hope is that if AI could be trained on the samples to identify "biomarkers", it would help get people diagnosed sooner, he said.

"It will help reduce false positives, it will help reduce false negatives, and hopefully reduce waiting times for people for colonoscopies and get people on the proper treatment trajectory."

Currently, once an FIT test had been done, the sample was discarded.

"They go into a bucket and then removed. So the idea is to [use] those discarded tests and then work out what bacteria are present in the tubes and train an AI to distinguish between the people who have bowel cancer versus those that are healthy, who might have blood in their stool for a number of other reasons.

"So it's really augmenting the existing screening platform that the National Bowel Screening Programme has.

"The tricky thing ... is sometimes it's not just the case that one bacteria equals increased likelihood. Sometimes you might have an increase in one or a decrease in another set of protective species or even more complicated relationships.

"Now, AI is the only tool in our scientific toolbox that's able to model these without having anything predefined. And so this is why it is incredibly exciting in preventative medicine that this is now being used in research and [we are] just trying to find some funding now to get this project off the ground."

There was strong international interest "in this space" but New Zealand's screening programme gave it an edge.

"The thing about New Zealand is we have this symptomatic screening programme which is basically anyone who presents - it's likely to be rolled out next year - people who present with bowel symptoms, they ... will be allowed to use this test to triage them for a colonoscopy. It'll put them on a certain list and that basically will help assess getting people the colonoscopy when they need to get it.

"The other thing is that New Zealand has the fastest rate of early onset bowel cancer in the world. And so ... there's an urgent need to get this research done quickly to help the future generations."

Why New Zealand had the fastest rate was unknown.

"Nobody knows why this is happening. Nobody knows why it's increasing. But there are certain risk factors that increase the odds.

"People are looking at microplastics. I'm interested in nitrates, how nitrates might fertilise the gut microbiome to increase the risk of the disease.

"But there are also lifestyle risk factors as well ... things like your BMI and smoking, obesity and of course age. But there's no definitive reason why early onset is increasing so much quicker here than anywhere else.