Scientists Are Using Artificial Intelligence to Develop New Drugs Using Anonymous Data

Monday, 22 October 2018 - 12:09PM
Technology
Medical Tech
Artificial Intelligence
Monday, 22 October 2018 - 12:09PM
Scientists Are Using Artificial Intelligence to Develop New Drugs Using Anonymous Data
< >
Composite from Unsplash/Pixabay
If there's one thing drug companies hate more than unforeseen side effects, it's sharing proprietary information about their drugs. Fortunately, a new study published in Science has shown that pharmaceutical companies can contribute to the ongoing process of discovering new drugs (and new uses for them) by training artificial intelligence on their data, which only the AIs are able to access.

One of the most complex aspects of testing for a new drug is understanding how it will interact with the body's proteins. Because the number of potential interactions for any given drug is often astronomically high, it's possible for humans to miss potential reactions that might be helpful...or ones that might trigger damaging side effects. Artificial intelligence, however, has proven that it's really, really good at trawling through huge reams of data and answering basic questions. The only issue is that a well-trained neural net (a type of artificial intelligence that can learn from experience) needs a lot of data to train itself.

This is where the pharmaceutical companies come in. They have a huge amount of data on drugs and various interactions, and by encrypting all that data on various servers they can share all of it with researchers so that the humans can't read it, but the neural nets can. So how do you prove that this black box of data is effective? Well, after training their neural net on the data provided to them, researchers Bonnie Berger, Brian Hie, and Hyunghoon Cho tested their AI by giving it millions of drug-protein pairs to sort through and identify. By the end of it, the AI was able to pick out which pairs interacted with 95% accuracy, discovered previously unknown interactions, and identified known interactions that weren't listed on the database the researchers were using.

According to Jian Peng, from the University of Illinois at Urbana-Champaign: "This work is visionary. I think [it] will lay the groundwork for the future of collaborations in biomedicine."
Science
Science News
Technology
Medical Tech
Artificial Intelligence
No