Artificial intelligence has learned to accurately detect celiac disease

Traditionally, diagnosis requires a duodenal biopsy, which is then studied by a pathologist. However, this method does not always provide unambiguous results, as tissue changes may not be obvious and their interpretation may be subjective. The new tool, trained on thousands of samples, analyzes the images and classifies them according to an accepted scale, which significantly increases the accuracy of diagnosis.

Tests have shown that the algorithm is able to correctly identify the presence or absence of disease in 97% of cases, and its accuracy is comparable to the conclusion of an experienced pathologist. Moreover, AI proved to be as reliable as a consilium of specialists: when comparing the diagnoses of doctors and the algorithm, it turned out that the level of coincidence between them was the same.

The use of such technologies can significantly speed up diagnosis, especially in regions where there is a shortage of specialists. Patients with celiac disease often face a lengthy diagnostic process that can take years. Using AI will shorten this timeframe and allow treatment to be administered more quickly, with doctors focusing on more complex cases.

The researchers’ immediate plans include large-scale clinical trials and certification of the technology for implementation in medical practice. If the algorithm gets approval, it could be used in hospitals, making diagnosis easier and helping thousands of patients get the right treatment faster.