Researchers have developed an artificial intelligence (AI) tool that can accurately diagnose acute otitis media (AOM), or ear infection, in children simply by analyzing a quick smartphone video of the eardrum. This new technology could revolutionize diagnosing and treating this common childhood condition.
The big challenge: While very common, accurately diagnosing AOM in squirming infants and toddlers requires intensive clinical training to catch subtle eardrum abnormalities in a brief examination window.
Why it matters
- AOM is frequently misdiagnosed, leading to either inadequate treatment or unnecessary antibiotic prescriptions that fuel antibiotic resistance.
- AOM is frequently confused with non-bacterial "otitis media with effusion" — fluid behind the eardrum not requiring antibiotics.
- An accurate diagnosis is crucial for providing proper care and avoiding overprescribing antibiotics.
Details
A team from the University of Pittsburgh Medical Center and the University of Pittsburgh created the smartphone app. It uses advanced AI models trained to identify visual signs of AOM on eardrum videos.
How it works:
- The researchers built a training library of over 1,100 eardrum video clips taken during pediatric visits from 2018-2023.
- Experts annotated each video, marking whether AOM was present based on specific visual criteria like eardrum bulging and color.
"In AOM, the eardrum bulges outward like an inflated bagel, leaving a central depression resembling a bagel hole," explained Dr. Alejandro Hoberman, the study's senior author. "This bulging is a key visual indicator we trained the AI to identify."
By the numbers:
- When tested on new videos, the two AI models showed over 93% accuracy in diagnosing AOM.
- Previous studies found human clinician diagnostic accuracy ranging from only 30-84%.
Yes, but: AI is not intended to replace clinicians but to assist them in making more accurate diagnoses. Dr. Hoberman noted, "Our tool helps get the correct diagnosis and guide the right treatment."
The takeaway
With over 93% diagnostic accuracy, the new AI tool outperforms most clinicians. If widely adopted in primary care, it could enhance AOM diagnosis, guide proper antibiotic prescribing, and help preserve antibiotic effectiveness long-term
- An added benefit: The eardrum videos can be securely stored in patients' records to document findings, view them with parents, and use to train medical students.
