AI tool developed by Cambridge researchers could accelerate coeliac diagnosis
Pathology slides: non-coeliac (left) and coeliac (right). Credit: University of Cambridge.
27 March 2025
An AI (artificial intelligence) tool developed by NIHR Cambridge Biomedical Campus researchers could dramatically increase the speed of diagnosis for people with coeliac disease. By automating the analysis of biopsy images, this could allow samples to be processed faster, reduce waiting lists and enable specialist pathologists to aid the diagnosis and treatment of more people.
Coeliac disease affects around 1 in 100 people who experience symptoms including stomach cramps, diarrhoea, skin rashes, weight loss, fatigue and anaemia as a result of eating gluten.
Diagnosis of coeliac disease is done by specially trained pathologists, who analyse biopsy samples taken from the intestine. This can be a time-consuming process, and it is not always possible to achieve a conclusive result.
“Anything that makes the system quicker must be a good thing, because once you’ve been diagnosed and you know you can’t have gluten, then you know what to do.”
Liz Cox, age 80, living with coeliac disease
A study published today (27 March 2025) in the New England Journal of Medicine, shows how AI could accelerate this process and take pressure off healthcare resources while providing comparable levels of accuracy. It could also offer solutions in developing nations that have little or no access to specialist pathologist.
While AI has already been widely investigated for its ability to detect cancer cells in biopsies, this is the first time that the same has been attempted for coeliac disease.
“Coeliac disease affects as many as one in 100 people and can cause serious illness, but getting a diagnosis is not straightforward. It can take many years to receive an accurate diagnosis, and at a time of intense pressures on healthcare systems, these delays are likely to continue. AI has the potential to speed up this process, allowing patients to receive a diagnosis faster, while at the same time taking pressure off NHS waiting lists.”
Professor Liz Soilleux, study lead and Honorary Consultant Pathologist
In the study, a type of AI known as a machine learning algorithm was shown 4,000 sets of biopsy images collected from five NHS hospitals to help it distinguish between healthy samples and those with coeliac disease.
When the AI was put to the test with another 650 biopsies, it made a correct diagnosis in more than 97 out of 100 cases when compared to diagnoses made by human pathologists. This is a landmark achievement since previous work by the same team has shown that human pathologists frequently disagree over diagnoses.
“This is the first time AI has been shown to diagnose as accurately as an experienced pathologist whether an individual has coeliac or not. Because we trained it on data sets generated under a number of different conditions, we know that it should be able to work in a wide range of settings, where biopsies are processed and imaged differently.
“Our next step is to test the algorithm in a much larger clinical sample, putting us in a position to share this device with the regulator, bringing us nearer to this tool being used on the NHS.”
Dr Florian Jaeckle from the Department of Pathology and first author on the paper
The study is led by Professor Liz Soilleux, Honorary Consultant Pathologist at Cambridge University Hospitals NHS Foundation Trust (CUH) and Professor of Diagnostics & Biomarkers at the University of Cambridge.
Find out how you can take part in research at CUH with Love Research
The researchers have been working with patient groups, including through Coeliac UK, to understand the patient response to AI-aided diagnosis.
A key concern they have identified from patients and clinicians that they are seeking to address is ‘explainability’. AI is able to come to conclusions based on patterns it sees in the data it is shown, but at the moment it’s not always easy for a human to understand what those patterns are, or to see whether they are actually relevant to the disease being diagnosed.
Creating AI diagnoses that are explainable is key for the researchers and is likely to be a critical step in the AI being approved, and trusted, for use across the NHS.
The research was funded by the National Institute for Health and Care Research. Lyzeum Ltd is a spinout company from the University of Cambridge that has been setup to commercialise this AI tool.
