Paging Doctor AI

Every year, millions of people worldwide are affected by some kind of lung disease, usually diagnosed with chest X-rays. Unfortunately, there aren’t enough expert radiologists to keep up with analysing all these images, particularly in areas of the world where doctors are in short supply and lack specialist training. As a solution, researchers have developed an artificial intelligence (AI) tool called CheXNeXt to analyse chest X-rays and spot the signs of 14 different lung diseases. For example, the images on the right show brightly coloured ‘heat maps’ that indicate the presence of abnormalities where vein catheters have been inserted (top) while the lower panels are from a patient with pneumonia, which shows up as a white mass on the X-ray. When put to the test, CheXNeXt performs at least as well as expert radiologists, so it could be a useful tool for helping to diagnose lung disease in the future.

Written by Kat Arney

Kat Arney

Writer

Dr Kat Arney is an award-winning science writer, broadcaster and public speaker, and is the founder and Creative Director of First Create The Media.

Website

• Image adapted from work by Pranav Rajpurkar and Jeremy Irvin, and colleagues
• Department of Computer Science, Stanford University, Stanford, CA, USA
• Image originally published under a Creative Commons Licence (BY 4.0)
• Published in PLOS Medicine, November 2018