Skip to Content
Artificial intelligence

DeepMind’s algorithm can predict serious kidney injury 48 hours before it happens

August 1, 2019
Dialysis
DialysisAP

It’s only accurate a little more than half the time, though, raising questions over whether and how it could be used in real-life settings.

The news: DeepMind, Google’s artificial-intelligence unit, has created a deep-learning system that analyzes someone’s health records, including information such as vital signs and blood test results, and then predicts the chance of acute kidney injury (AKI). In a paper published in the journal Nature yesterday, researchers showed they were able to predict 55.8% of cases up to 48 hours before they occurred. For more severe kidney injuries, such as those that would require dialysis at a later stage, the accuracy was closer to 90%. 

The promise: AKI contributes to almost 300,000 deaths in the US every year, and affects one out of every five patients admitted to hospital in America for serious care. However, if it’s caught and treated early, it can be reversed. That’s why this system is so promising: it could potentially save a lot of lives.

However: It was built using records from the US Department of Veterans Affairs, a training data set that was 94% male. It would need a lot more testing to confirm its utility in the general population. There were also many false positives.

DeepMind plans to incorporate the algorithm into its Streams app, which helps doctors at London’s Royal Free Hospital to identify patients at risk of developing AKI. Only then, after further testing, can we say for sure just how useful it will be.

For more on the world of AI, sign up here to our weekly AI newsletter, The Algorithm.

Deep Dive

Artificial intelligence

What does GPT-3 “know” about me? 

Large language models are trained on troves of personal data hoovered from the internet. So I wanted to know: What does it have on me?

An AI that can design new proteins could help unlock new cures and materials 

The machine-learning tool could help researchers discover entirely new proteins not yet known to science.

DeepMind’s new chatbot uses Google searches plus humans to give better answers

The lab trained a chatbot to learn from human feedback and search the internet for information to support its claims.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.