Watch CBS News

"Gamechanger" AI smartphone app developed in Pittsburgh can diagnose ear infections more accurately than many clinicians

Pittsburgh physicians develop tool that uses AI to better detect ear infections
Pittsburgh physicians develop tool that uses AI to better detect ear infections 02:11

PITTSBURGH (KDKA) — A new smartphone app developed by researchers at UPMC and the University of Pittsburgh uses artificial intelligence to accurately diagnose ear infections. 

The AI tool analyzes short videos of the eardrum that are captured by a smartphone-connected otoscope.

"We accomplished 94% sensitivity specificity and 94% accuracy in diagnosing an ear infection versus not an ear infection in children, which is quite good and better than most physicians are," said Alejandro Hoberman, director of the Division of General Academic Pediatrics at Pitt's School of Medicine and president of UPMC Children's Community Pediatrics.

Dr. Hoberman has been studying acute otitis media for decades. He said past studies of clinicians have reported diagnostic accuracy of acute otitis media ranging from 30 to 84 percent.

Accurately diagnosing acute otitis media requires a trained eye to pick up subtle visual findings from a quick look at the eardrum of a squirming baby. It's also often confused with another condition that generally doesn't involve bacteria and doesn't benefit from antimicrobial treatment, researchers say. 

Hoberman said around 70 percent of children get this ear infection before their first birthday, but it's often misidentified.

"It hurts, and the quality of life is not good. And the kids wake up at night and they're uncomfortable and they're fussy," he said.

Hoberman and his team used over 900 videos from a training library they created to teach two different AI models to detect acute otitis media, then used about 200 videos to test how the models performed. The AI tool uses short videos captured on an otoscope that's connected to a smartphone.

"We are talking to the software and actually telling capture, stop and then we trim it to like two or three or four seconds," Hoberman said. "This is not generating AI. This is not inventing anything. This is using the way we think as pediatricians and what's written."

Hoberman said the infection is the number one reason children are prescribed antibiotics. According to new research published in JAMA Pediatrics, the groundbreaking tool Hoberman and his team developed could lower unnecessary antibiotic use in kids.  

This is a big win for physicians because Hoberman said ear infections can be both underdiagnosed and overdiagnosed.  

"If they're overdiagnosed, the child is getting inadequate care. They're getting unnecessarily antibiotics. And if they are not diagnosed, if they're underdiagnosed, the child is not getting the right care," he said.

The hope is that the technology will soon be certified by the FDA and used across doctors' offices.

"We love it, in using it, in teaching it and sharing with parents in shared decision-making, in enhancing the quality of their observations. We believe it's going to be a gamechanger. And hopefully, there will be adoption of this, it'll become a product that people will be able to download and use on an easy basis," Hoberman said.

Hoberman said the otoscope they use is something anyone can buy online, which means the app could potentially be used by parents to send video files to pediatricians. They're using the tool at 10 pediatric offices for clinical research.

"Eventually, these videos will be able to be incorporated into the electronic health record. And for the purposes of researching new infections, we are requiring it and the FDA actually requires that places that are participating in Otitis Media Research actually get video images of the drum," Hoberman said.

View CBS News In
CBS News App Open
Chrome Safari Continue
Be the first to know
Get browser notifications for breaking news, live events, and exclusive reporting.