App uses AI to identify COVID-19

POCATELLO, Idaho (KIFI/KIDK) - How would you feel if you could get coronavirus test results on your phone in a matter of seconds? Well, you might be able to soon.

Utah-based AI Biokinetic Technologies, with Aspen Clinical Research, is in the process of getting final approval on an app that can give you coronavirus results. Using artificial intelligence, the COVID AI 19 app will give you a thermal temperature reading, an audio analysis and analyze an X-ray of your lungs, all on your device. 

"Get the X-ray done and then digitally upload and then boom! You have a result very rapidly, as soon as we got that X-ray or CT scan uploaded into the AI," Wayne Croft, COO of AI Biokinetic Technologies, said. "Within four seconds."

All you'll need to do (when the app is released) is open your phone, sign a consent form, fill out some basic info, do a face scan and then upload a picture of your chest X-ray.

The technology was developed several years ago by Idaho natives Zach Newman and Kade France. In the time since then, their system has analyzed hundreds of thousands of images to identify common respiratory illnesses.

When the group found out that coronavirus was detectable through chest X-rays, they shifted focus.

“It took a matter of only a few days - a couple days- to enable it to detect the coronavirus," France, the project's lead software engineer, said.  

Comparing lab results with more than 40,000 COVID-19 scans from places like China, Italy and Spain, the group says the system has been highly successful so far. 

"We were seeing that within about 93-96% accuracy on the scans that were available," Chief Medical Office Ian Dawe said. "And one of the components of artificial intelligence and using it is that it continues to become more accurate as we continue to implement and add more data to it."

Using AI would allow these tests to be conducted anywhere an X-ray could be done. Keeping people from having to come in contact with each other could also reduce potential exposures.

"Instead of going to a hospital and trying to set up at a hospital, we could go to a high school or a church community or a mall setting and set up a tent," Dawe said of an effort to "clear" people.

The technology is out there and ready to be deployed to test and clear individuals. Once approved by the ethics board, the group plans to run clinical trials, hoping to get about 500 individual scans before they can submit results to the FDA for approval. 

“We don’t have to see all new patients, we can look at patients who have already been there, our system just needs to analyze the data," Daw said, explaining that anyone who was tested and then followed up with an X-ray can submit data.

"We’ll be able to put those two things together, that and the results from the data scan, and make sure that what we’re seeing is accurate and gives it that confidence we need. We want to get this out in the community as quickly as possible."

Patients all around the United States can submit lab and scan data if they've been previously tested for COVID-19.

"We can take that data and put it into our machine learning and that would help us to further validate it," Croft said. "We can do it prospectively and retroactively."

The group said they just need data to finish the process. They hope to see a 96% probability of accuracy during clinical trials. 

The app was approved by the Apple App Store on Thursday. The group is currently working on an Andriod version.