Deep learning, an aspect of artificial intelligence (AI), has recently been shown to interpret clinical images more accurately than humans — but it’s unlikely to replace rheumatologists and radiologists anytime soon, according to a review.

A leading authority on rheumatological imaging, Associate Professor Berend Stoel, says that, in the short term, it’s more likely that AI will become an adjunct to human reading of images.

Professor Stoel, from the Division of Image Processing at Leiden University Medical Center, the Netherlands, says AI could be useful for recognising patterns among huge sets of clinical data, including images that the human eye and brain cannot encompass.

“Deep learning is based on the concept of artificial neural networks that mimic human learning capacity using mathematical representations of neurons and their interconnections,” he says.