AI won’t replace doctors – but it could help us prevent cancer
Professor Colin Rees explains Newcastle University’s groundbreaking Colo-Detect trial.
A picture of health: The positive impact artificial intelligence can have in healthcare
How can artificial intelligence help diagnose cancer? In this blog, Professor Colin Rees explains Newcastle University’s groundbreaking COLO-DETECT trial, offering AI-assisted colonoscopies to patients and helping clinicians spot precancerous lesions sooner.
For many people, discussions about AI focus on uncertainty: what jobs it might replace, what decisions it should be allowed to make, and what role it should play in our daily lives. But in healthcare, AI has the potential to help clinicians find disease earlier, make better decisions and ultimately save lives.
In this blog, Newcastle alumnus Professor Colin Rees, President of the British Society of Gastroenterology from 2024-26, explains how AI-assisted colonoscopy is improving the early detection of bowel cancer.
The challenge we face
Bowel cancer remains one of the biggest cancer challenges worldwide. It is the third most common cancer globally and the second biggest cause of cancer-related death. In the UK alone, around 42,000 people are diagnosed with bowel cancer each year, and more than 16,000 lose their lives to the disease.
Here in the North East, outcomes are particularly concerning. Our region continues to experience some of the highest rates of bowel cancer and some of the poorest outcomes.
Yet there is an important reason to be optimistic. Many bowel cancers are preventable. In fact, bowel cancer usually develops slowly over many years.
Most cancers begin as small growths called polyps. These can take a decade or more to become cancerous, creating a valuable opportunity to identify and remove them before cancer develops. The challenge is finding every polyp.
Why detection matters
Colonoscopy is currently our best tool for preventing bowel cancer. During a colonoscopy, a specialist examines the lining of the bowel looking for abnormalities. If polyps are found, they can often be removed immediately.
Research has consistently shown that the more precancerous polyps we detect and remove, the fewer patients go on to develop bowel cancer.
But colonoscopy is performed by humans, and humans are not perfect. Even highly experienced specialists can occasionally miss small or difficult-to-spot lesions. Some polyps are tiny. Others are flat and blend into the surrounding bowel lining. Some are located in areas that can be particularly challenging to identify.
This is where AI can make a difference. Not by replacing clinicians, but by supporting them.
Colonoscopy is performed by humans, and humans are not perfect. Even highly experienced specialists can occasionally miss small or difficult-to-spot lesions. This is where AI can make a difference.
A second set of eyes
The type of AI we are using in bowel cancer detection is based on machine learning. By analysing thousands of images of normal and abnormal bowel tissue, AI systems learn to recognise patterns that may indicate a polyp. During a colonoscopy, the software can then highlight suspicious areas in real time, effectively providing a second set of eyes for the endoscopist.
At Newcastle University, we wanted to understand whether this technology could genuinely improve outcomes in everyday clinical practice. That question led to the COLO-DETECT trial. The trial involved nearly 2000 patients across ten NHS trusts in England. Participants were randomly assigned to undergo either a standard colonoscopy or a colonoscopy supported by an AI system known as GI Genius.
The results were striking. We found that using AI significantly increased the number of adenomas, the precancerous polyps most likely to develop into bowel cancer, that were detected.
Overall, AI-assisted colonoscopy increased adenoma detection by more than a third compared with standard procedures. Importantly, it was particularly effective at identifying smaller and flatter lesions, exactly the types of polyps most likely to be overlooked.
We also saw improvements in the detection of sessile serrated lesions, a particularly challenging type of polyp that is responsible for a disproportionately large number of bowel cancers.
The procedures took slightly longer, which is unsurprising if more abnormalities are being found and removed. Crucially, there was no increase in complications or adverse events. In short, the technology worked.
AI significantly increased the number of adenomas, the precancerous polyps most likely to develop into bowel cancer, that were detected.
Why this matters for patients
The significance of these findings extends far beyond a single clinical trial. Decades of research have demonstrated a clear relationship between adenoma detection and future cancer risk. Higher detection rates are associated with fewer cancers developing after colonoscopy and, ultimately, fewer cancer-related deaths.
The evidence suggests that the improvements we observed could potentially reduce post-colonoscopy bowel cancers by around 25%.
For patients, that is not a technical statistic. It represents lives saved, cancers prevented, and families spared the devastating impact of a cancer diagnosis.
The future of AI in healthcare
What excites me most is that this is only the beginning. AI is already helping healthcare professionals analyse scans, prioritise referrals, identify patterns in large datasets and support clinical decision-making. In bowel cancer, we are also exploring how AI can improve screening pathways, helping us identify which patients are most likely to benefit from further investigation.
The healthcare systems of the future will face growing pressures from ageing populations, increasing demand and workforce shortages. AI will not solve these challenges on its own. Nor should it replace the expertise, judgement and compassion that clinicians bring to patient care.
But it can help us work smarter. It can reduce variation. It can support decision-making. And it can help ensure that more patients receive the right investigation at the right time.
The evidence suggests that the improvements we observed could potentially reduce post-colonoscopy bowel cancers by around 25%.
Innovation with patients at its heart
One lesson from the COLO-DETECT trial stands out above all others: none of this would have been possible without patients.
Every advance in healthcare depends on people willing to participate in research and help us answer important questions. Their contribution allows new technologies to move from exciting ideas to real-world solutions.
At Newcastle University, our ambition is clear: to continue leading research that improves cancer prevention, diagnosis and treatment. The Colo-Detect trial shows what is possible when clinicians, researchers, patients and technology work together.
AI will not replace doctors. But if it helps us find more cancers earlier, prevent more deaths and give more people the chance of a healthier future, then it is a tool we should embrace.
As health systems strain under rising demand, workforce shortages, and widening inequities, AI promises breakthroughs in diagnostics, drug discovery, and operational efficiency.
Yet the same technologies raise urgent concerns around safety, bias, accountability, and the future of clinical work.
Join our panel of Newcastle University alumni and colleagues to explore the future of healthcare through the lens of AI’s most transformative and most troubling possibilities.
This event is brought to you in collaboration with our INSIGHTS Public Lectures programme.
