As coronavirus AI algorithms continue to roll out, some radiologists are urging caution

Studies continue to trickle out touting the ability of artificial intelligence to pinpoint coronavirus on chest images, and technology companies have pivoted to capitalize on creating such algorithms. Skeptics, including radiologists, however, are starting to question the validity of these platforms.

STAT news spoke with both developers and imaging experts about this topic for its March 28 story. The latter raised a number of questions, including concerns that AI algorithms are not being trained on diverse and well-labeled datasets, and that images used focus heavily on critically sick patients.

“I’ve looked at the work around AI and COVID, and I’m not a big fan,” Ella Kazerooni, MD, a radiologist at the University of Michigan Medical Center, told the news outlet. “When you’re just comparing normal and abnormal scans, it’s not very helpful.”

Developers, such as Moshe Becker, chief executive of RadLogics, which recently released a new COVID-19 AI platform, argue that their approaches are accurate and are already being used in clinical settings.

“We have processed hundreds of additional patient cases in China and Russia, and we have started in Italy … and we are seeing similar results and performance,” Becker told STAT.

Read the entire story below.