A new mobile phone app developed by UPMC and University of Pittsburgh physician-scientists uses artificial intelligence (AI) to accurately diagnose ear infections or acute otitis media (AOM) and eliminate unnecessary antibiotics in young children. New research released today could help reduce substance use JAMA Pediatrics.
AOM is one of the most common pediatric infections for which antibiotics are prescribed, but without intensive training it is difficult to distinguish it from other ear conditions. This new AI tool makes a diagnosis by evaluating a short video of the eardrum taken with an otoscope connected to a mobile phone camera, and is simple and effective, potentially more accurate than a trained clinician. We provide practical solutions.
Acute otitis media is often misdiagnosed. Underdiagnosis can result in inappropriate treatment, and overdiagnosis can result in unnecessary antibiotic treatment, compromising the effectiveness of currently available antibiotics. Our tools will help you get the correct diagnosis and guide the appropriate treatment. ”
Alejandro Hoberman, MD, senior author, professor of pediatrics, chair of general academic pediatrics at Pitt School of Medicine, and chairman of UPMC Children's Community Pediatrics
About 70 percent of children have an ear infection by their first birthday, Hoberman said. Although this condition is common, accurately diagnosing AOM requires a trained eye to detect subtle visual findings that can be obtained by simply observing a baby's wriggling eardrum. AOM is often confused with otitis media, which is an accumulation of effusion or fluid behind the ear, but this condition generally does not involve bacteria and does not benefit from antimicrobial treatment.
To develop a practical tool to improve the accuracy of diagnosing AOM, Hoberman and his team analyzed 1,151 tympanic membranes from 635 children who visited the UPMC pediatric outpatient clinic between 2018 and 2023. I started by building and annotating a training library of videos for the book. His two trained experts with extensive experience in AOM research reviewed the video and diagnosed whether he had AOM or not.
“The tympanic membrane, or tympanic membrane, is a thin, flat piece of tissue that spans the entire ear canal,” Hoberman says. “In AOM, the eardrum bulges like a bagel, leaving a depressed central area that resembles the hole in a bagel. In contrast, in children with otitis media with effusion, the tympanic membrane bulge is not present.”
The researchers used 921 videos from a training library to teach two different AI models to detect AOM by examining eardrum characteristics such as shape, location, color, and translucency. We then used the remaining 230 videos to test the performance of our model.
Both models were highly accurate, yielding sensitivity and specificity values above 93%. This means that the rate of false negatives and false positives is low. Previous studies of clinicians have shown that the diagnostic accuracy of AOM ranges from 30% to 84%, depending on the type of provider, level of training, and age of the child being tested, Hoberman said. It is reported that there is.
“These findings suggest that our tool is more accurate than many clinicians,” Hoberman said. “Supporting clinicians to rigorously diagnose her AOM and make treatment decisions could be a game-changer in primary health care settings.”
“Another benefit of our tool is that the video captured can be saved in the patient's medical record and shared with other healthcare providers,” Hoberman said. “We can also show parents, residents, medical students and residents what we see and explain why ear infections are being diagnosed or not diagnosed. This is important as an educational tool and to reassure parents that their child has an ear infection and appropriate treatment. ”
Hoberman hopes his technology will soon be widely deployed across healthcare provider offices to enhance accurate diagnosis of AOM and support treatment decisions.
Other authors of the study were Nader Shaikh, MD, Shannon Conway, MD, Timothy Shope, MD, Mary Ann Harlam, MD, CRNP, Catherine Campese, MD, and Matthew Lee, MD, of UPMC and the University of Pittsburgh. Dr. Jelena Kovačević of New York University; Dr. Filipe Condesa of the Bosch Center for Artificial Intelligence. and his Tomas Larsson, MSc, Zafer Cavdar of Dcipher Analytics.
This research was supported by the University of Pittsburgh School of Medicine Department of Pediatrics.
sauce:
Reference magazines:
Sheikh, N. other. (2024). Development and validation of an automatic classifier for diagnosing acute otitis media in children. JAMA Pediatrics. doi.org/10.1001/jamapediatrics.2024.0011.