For any parent who has endured the misery of a young child with an ear infection —and with most children expected to have one before age 3, there are plenty of you out there — here is welcome news.
Researchers at Massachusetts Eye and Ear in Boston have created an artificial intelligence-based model that outperformed doctors, including specialists and pediatricians, in accurately diagnosing ear infections.
The model, called OtoDX, was more than 95% accurate than clinicians in diagnosing an ear infection in a set of images. Shown the same images, the doctors accurately diagnosed the infections about 65% of the time.
Diagnosing ear infections, especially in young children, can be tough.
Babies and young children who aren’t feeling well often squirm or cry and aren’t particularly agreeable to staying put long enough for a doctor to get a good look.
But accurately diagnosing ear infections is critical. Untreated, they can cause hearing loss and developmental delays. And prescribing antibiotics when there is no infection leads to antibiotic resistance, a global health problem.
The model now has more than 1,000 images of tympanic [tim-panic] membranes from children under 18 who were set to undergo tube-placement surgery or have ear fluid drained. The tympanic membrane, better known as the eardrum, separates the outer and middle ear.
Studies are underway to validate the model. In addition, OtoDX is being tested in a prototype device that pairs with a smartphone to allow clinicians to quickly snap photos that can be quickly analyzed.
In short: The time is coming when little ones’ ear issues can be diagnosed more quickly and accurately. Music to everyone’s ears.