AI tool created by University of Cambridge scientists is as good at pathologists at diagnosing coeliac disease

An artificial intelligence tool developed by Cambridge scientists has proved to be as good as pathologists at diagnosing coeliac disease.

The machine learning algorithm correctly identified in 97 cases out of 100 whether or not an individual had the disease, based on their biopsies.

The tool has been trained on almost 3,400 scanned biopsies from five hospitals and could speed up diagnosis of the condition and reduce pressure on healthcare resources.

In developing nations, where there are severe shortages of pathologists, it could improve diagnosis.

Prof Elizabeth Soilleux, from the Department of Pathology and Churchill College at the University of Cambridge, who is senior author of research on it, said: “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.”

Coeliac disease is an autoimmune disease, triggered by consuming gluten, which causes symptoms such as stomach cramps, diarrhoea, skin rashes, weight loss, fatigue and anaemia.

The variation in symptoms means patients often find it difficult to get an accurate diagnosis.

A biopsy of the duodenum - part of the small intestine - is the gold standard for diagnosis. Pathologists analyse samples under a microscope or on a computer to look for damage to the villi, which are tiny hair-like projections that line the inside of the small intestine.

But interpreting subtle changes in biopsies is not easy, and pathologists use the Marsh-Oberhuber scale to judge the severity of a case, ranging from zero - in which the villi are normal and the patient is unlikely to have the disease - to four, which means the villi are completely flattened.

The new research, published in the New England Journal of Medicine AI, explains how researchers tested their algorithm on an independent data set of almost 650 images from a previously unseen source and, compared to the original pathologists’ diagnoses, they found the model was correct in its diagnosis in more than 97 cases out of 100.

The tool had a sensitivity of more than 95 per cent, meaning it correctly identified more than 95 cases out of 100 individuals who had coeliac disease, and had a specificity of almost 98 per cent, meaning it correctly identified nearly 98 cases out of 100 individuals who did not have coeliac disease.

Pathologists disagree on diagnosis in more than one in five coeliac disease cases, so the researchers asked four pathologists to review 30 slides and found they were as likely to agree with the AI model as they were a second pathologist.

Dr Florian Jaeckle, also from the Department of Pathology, and a research fellow at Hughes Hall, Cambridge, said: “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.

“This is an important step towards speeding up diagnoses and freeing up pathologists’ time to focus on more complex or urgent cases. 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 in the NHS.”

The researchers worked with patient groups, including through Coeliac UK.

“When we speak to patients, they are generally very receptive to the use of AI for diagnosing coeliac disease,” said Dr Jaeckle. “This no doubt partly reflects their experiences of the difficulties and delays in receiving a diagnosis.

“One issue that comes up frequently with both patients and clinicians is the issue of ‘explainability’ – being able to understand and explain how AI reaches its diagnosis. It’s important for us as researchers and for regulators to bear this mind if we want to ensure there is public trust in applications of AI in medicine.”

The algorithm is now being commercialised by spin-out company Lyzeum Ltd, set up by Dr Jaeckle with Prof Soilleux, a consultant haematopathologist at Cambridge University Hospitals NHS Foundation Trust.

Patient Liz Cox, 80, who had symptoms including anaemia and stomach pains for almost 30 years before she went back to the doctors and got a diagnosis.

She said: “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, and you feel so much better.”

The research was funded by Coeliac UK, Innovate UK, the Cambridge Centre for Data-Driven Discovery and the National Institute for Health and Care Research.