Medical app uses AI to help predict coronavirus outbreaks

Diagnostic Robotics CTO Kira Radinsky joins Yahoo Finance’s Kristin Myers to discuss how the company is using AI to predict how fast the coronavirus will spread.

Video Transcript

KRISTIN MYERS: But as we were chatting about a little bit earlier, coronavirus continues to spread across the world. But what if we could predict exactly where, when, and how fast it could spread? Well, a new app from Diagnostic Robotics says that they might be able to do just that.

And we're joined now by Kira Radinsky. She's the chief technology officer for Diagnostic Robotics. Thanks so much for joining us, Kira.

KIRA RADINSKY: Thanks for having me.

KRISTIN MYERS: So Kira, I want to know just first how this app works. And then if you could chat with us a little bit about the accuracy of it.

KIRA RADINSKY: Definitely. So Diagnostic Robotics is a clinical triage system that has been working for approximately 50 million people across the globe till today with more than 100,000 different medical providers using that. It's not an app. It's actually an AI platform integrating different health care systems. Three months ago, we've been requested by the Israeli minister of health to not only triage for all the diseases but this time triage on a daily basis all the 8 million patients in Israel.

Using this clinical triaging, we've been able to estimate an epidemiological map showing where COVID is right now. But most importantly, using machine learning, we'll provide predictions about how it's going to spread. We were able to predict on April 4 three large cities, publish them on the median. And indeed, a few days later, the government has quarantined them based on our predictions. Since then on a daily basis, our report is being presented to the vice president of the Ministry of Health making any kind of clinical predictions and additionally quarantine strategy.

I must mention this is the only digital solution that Israel has deployed. And we've been out with that strategy for the last month.

The solution has also been monitoring 46 million people in [INAUDIBLE] India with our collaboration with Deloitte. And we recently launched in Rhode Island state as well.

KRISTIN MYERS: So Kira, I'm wondering if you can chat through. I know you mentioned this has been deployed in Israel. How successful or how accurate has it been in terms of predicting where the virus will go next?

KIRA RADINSKY: So our AUC is 82. So in other words, we rank cities. And if you look at the top 10, 82 on average will be correct ones based on the real actual ranking at the end.

KRISTIN MYERS: I'm wondering if you think that a tool like this could be successfully deployed here in the United States, especially as right now, in a moment of civil unrest, people are very cautious and very wary of technology having access, you know, to their phones and to their personal data. Do you see it being applicable here in the United States? I know you have a partnership with one state.

KIRA RADINSKY: Definitely. So we launched with Rhode Island. And not similar to the tracking information, our information is not keeping any PHI, personal information, at all. It's completely anonymous, aggregating only clinical information. This enabled us to learn a lot about the disease and its progress. We're one of the first companies to identify that loss of taste and smell correlates with the disease. For example. We also know that extreme fatigue combined with dry cough is much more correlative than just fever. What enables this is the collaboration of all of the people answering this on a daily basis, providing us insights about how the disease spreads.

KRISTIN MYERS: All right, Kira Radinsky, chief technology officer of Diagnostic Robotics. Thanks so much for joining us.

KIRA RADINSKY: Thank you.