PHOTO: SHUTTERSTOCK

If you feel like you’re always taking your child to the doctor for a suspected ear infection, there’s good news on the horizon.

Physician-scientists at UPMC and the University of Pittsburgh have developed an app that uses AI to make a diagnosis by assessing a short video of the eardrum captured by an otoscope, the apparatus doctors use to look inside of patients’ ears, connected to a smartphone camera. The tool could help to decrease unnecessary antibiotic use in young children, according to research published this week in JAMA Pediatrics. 

“Our tool helps get the correct diagnosis and guide the right treatment,” said Dr. Alejandro Hoberman, senior author on the study and a professor of pediatrics and director of the division of general academic pediatrics at Pitt’s School of Medicine, in a statement. He also serves as president of UPMC Children’s Community Pediatrics; several CCP sites and other UPMC sites have already been using the tool.

Eventually, Hoberman says, the hope is that families will be able to buy an off-the-shelf product and obtain a video at home of their child’s ear to share with their pediatrician.

“Your pediatrician can decide if you need to bring him or her in and the video produces the artificial intelligence,” he says.

The tool can also be used when patients are seen at an urgent care center. A medical assistant could take a video of the child’s ear when they get their vital signs, and the doctor and parent can look at the video together. 

“It will give an answer whether it’s an ear infection or not an ear infection with a confidence level,” says Hoberman. “When the doctor comes in to examine the child, we’ll be able to see that information, look at the video of the eardrum and decide in conjunction with the parent whether there is an ear infection or not and what is the benefit of antibiotics.”

It’s often difficult for care providers to get an accurate diagnosis of ear infections, whether due to inexperience or wiggly babies. If a child is prescribed antibiotics but doesn’t really need them, it makes the medicine less effective the next time the child is prescribed them, Hoberman says. 

The study examined 1,151 videos from 635 children who visited outpatient UPMC pediatric offices between 2018 and 2023. Two trained experts reviewed the videos and made a diagnosis.

The researchers used 921 of those videos to teach two different AI models to detect ear infections and the remaining 230 videos to test how the models performed. Both models had accuracy rates of more than 93%. 

According to Hoberman, previous studies of clinicians have reported diagnostic accuracy ranging from 30% to 84%, depending on the type of health care provider, level of training and age of the children being examined. 

“These findings suggest that our tool is more accurate than many clinicians,” he said in a statement. “It could be a gamechanger in primary health care settings.”

According to the National Insitute of Deafness and Other Communication Disorders, ear infections are the No. 1 reason why parents bring their child to the doctor. It’s more common in youngsters because their immune systems are developing and because their eustachian tubes are smaller and more horizontal, making it more difficult for fluid to drain out of the ear. In fact, five out of six children will have at least one ear infection by their third birthday, according to the federal health agency.