Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

A.I. Harnessed to Diagnose Children’s Gut Diseases

Sana Syed

Sana Syed (University of Virginia)

17 June 2019. Medical researchers and data scientists are using artificial intelligence, or A.I., to diagnose children’s intestinal diseases found in the developing world. A team from University of Virginia in Charlottesville, with colleagues in the U.K., Pakistan, and Zambia, describe their techniques in the 14 June issue of the journal JAMA Network Open.

Researchers led by Virginia pediatrics professor and global health scientist Sana Syed, and Donald Brown, director of the university’s Data Science Institute, are seeking faster and more accurate techniques for diagnosing environmental enteric dysfunction, or EED, a disease widespread in among children in low- and middle-income countries, often from poor sanitation. EED is responsible for stunted growth, from reduced functioning of the small intestine, as well as subsequent cognitive and immune system disorders. Yet, the symptoms of EED are similar to celiac disease, an autoimmune disease also found in children where the ingestion of gluten in wheat, rye, and barley causes damage to the small intestine.

Syed, Brown, and colleagues set out to design a process to diagnose these intestinal diseases remotely, by analyzing images from biopsies of gut tissue from patients. The process required not only the detection of intestinal disorders, but also an accurate understanding of the exact disease discovered.

The researchers developed their process around an A.I. technique called a convolutional neural network that combines features of image analysis and machine learning. In a convolutional neural network, an algorithm dissects an image by layers to understand the features in the image. Different aspects of each layer discovered and analyzed by the algorithm are translated into data that the algorithm then uses to train its understanding of the problem being solved, with that understanding enhanced and refined as more images and data are encountered.

In this study, the researchers trained their convolutional neural network algorithm with images from intestinal biopsies. The biopsies were taken from 102 children with intestinal disorders, and a median age of 2.5 years in Pakistan, Zambia, and the U.S.: University of Virginia hospital in Charlottesville. The entire population was divided about equally, 52 to 48 percent, between boys and girls. Biopsies were also taken from another 42 children in Charlottesville with no intestinal disorders, for a total of 3,118 images.

The team compared results of the convolutional neural network algorithm to on-the-spot diagnoses by the children’s physicians at the participating hospitals. The results show the algorithm accurately returned a correct disease diagnosis, discriminating between EED and celiac disease, or healthy intestinal tissue, 93 percent of the time. The algorithm missed diagnosing the disease about 2 percent of the time. In addition, the algorithm found subtle changes in tissue cells that secrete mucus, important for the gut’s healthy functioning.

The researchers believe their technique can make it possible to diagnose children’s intestinal disorders remotely, with a faster turnaround of results. “There is so much poverty and such an unfair set of consequences,” says Syed in a university statement. “If we can use these cutting-edge technologies and ways of looking at data through data science, we can get answers faster and help these children sooner.”

More from Science & Enterprise:

*     *     *

Comments are closed.