Celiac disease, an autoimmune disorder triggered by gluten consumption, affects approximately 1 in 100 people and is often misdiagnosed due to varying symptoms.

This research represents the first successful attempt to apply AI for diagnosing celiac disease, achieving accuracy comparable to experienced pathologists.

Elizabeth Soilleux, a senior author of the study, emphasized that the AI tool could alleviate long waiting times for patients, allowing quicker diagnoses amid healthcare pressures.

Next steps include testing the AI algorithm on a larger clinical sample to facilitate potential regulatory approval for its use in NHS settings.

A personal story shared in the study highlights Liz Cox, who suffered from undiagnosed celiac disease for nearly 30 years, underscoring the need for faster diagnostic tools.

Current diagnosis involves trained pathologists analyzing biopsy samples from the intestine, a process that can be slow and inconclusive.

Researchers from the University of Cambridge have developed an AI tool that was trained on 4,000 biopsy images from five NHS hospitals, achieving over 97% accuracy in diagnosing celiac disease.

The research team is focused on making AI diagnoses explainable to ensure trust and understanding among patients and clinicians, which is crucial for adoption in healthcare settings.

The study received support from organizations including Coeliac UK and Innovate UK, emphasizing the need for investment in digital pathology and training for effective AI implementation.

The AI's diagnostic accuracy matched that of experienced pathologists, suggesting its readiness for varied clinical settings and potential implementation in the NHS.

The study suggests that AI could prioritize celiac disease tests that currently face delays due to more urgent cases like cancer screenings.

The algorithm demonstrated a sensitivity of over 95% and a specificity of almost 98%, outperforming traditional diagnostic methods that can lead to disagreements among pathologists.