Washington: Ann Artificial intelligence model Pediatricians were shown to be significantly more accurate than doctors in diagnosing the disease Ear infections In the first head-to-head evaluation of its kind. A tool called AutoDX was more than 95 percent accurate in making the diagnosis the ear Infection in a set of 22 test images compared to 65 percent accuracy in a group of 389 physicians who reviewed the same images.
The model, called OtoDX, was more than 95 percent accurate in diagnosing ear infections in a set of 22 test images, compared with 65 percent accuracy in a group of clinicians, according to a new study published Aug. 16 in Otolaryngology-Head and Neck Surgery. Included were ENTs, pediatricians, and primary care doctors, who reviewed the same images.
When tested on a dataset of more than 600 inner ear images, the AI model had a diagnostic accuracy of over 80 percent, representing a significant leap over the average accuracy of clinicians reported in the medical literature.
The model uses a type of AI called deep learning and was built from hundreds of photographs collected from children before undergoing surgery at Mass Eye and Ear for frequent ear infections or ear fluid. The results represent a major step toward the development of a diagnostic tool that could one day be deployed in clinics to assist doctors during patient evaluation, according to the authors. AI-based diagnostic tool providers, such as pediatricians and urgent care clinics, can offer additional tests to better inform their clinical decision-making.
“Ear infections in children are incredibly common yet frequently misdiagnosed, leading to delays in care or unnecessary antibiotic prescriptions,” said Matthew, lead author of the study. CrowsonMD, Otolaryngologist and artificial intelligence Mas Eye & Ear Researcher and Assistant Professor of Otolaryngology-Head and Neck Surgery at Harvard Medical School. “This model will not replace clinicians’ judgment but may complement their expertise and help them be more confident in their treatment decisions.”
A common condition is difficult to diagnose
Ear infections are caused by the accumulation of bacteria inside the middle ear. According to the National Institute on Deafness and Other Communication Disorders, at least five out of six children in the United States have at least one ear infection before the age of three. When left untreated, ear infections can lead to hearing loss, developmental delays, complications such as meningitis, and death in some developing countries.
Conversely, overtreating children when they don’t have an ear infection can lead to antibiotic resistance and make medications ineffective against future infections. This latter problem is important for public health.
To ensure the best outcomes for children, clinicians should diagnose ear infections as accurately and early as possible. However, previous studies suggest that the accuracy of conventional diagnosis of ear infections in children from physical examination is routinely below 70 percent, despite innovations in technology and clinical practice guidelines. Dr. According to Crosson, the difficulty in evaluating a child who is struggling or crying during the exam, along with the general inexperience many doctors and urgent care providers have in ear evaluations, may explain the lower-than-expected diagnosis rate.
“Since doctors prefer to err on the side of caution, it’s very easy to see why parents usually walk out of urgent care with a prescription for antibiotics,” he said.
In 2021, Dr. Crowson was joined by Mass Eye and Ear colleagues Michael S. Cohen, MD, Associate Director Multidisciplinary Pediatric Hearing Loss Clinicand Christopher J. Hartnick, MD, MS, Director Department of Pediatric Otolaryngology, to develop a more accurate method of diagnosing ear infections using machine learning algorithms. An artificial neural network was trained with high-resolution, tympanic membrane photographs collected directly from patients during ear procedures where infections could be seen. These photos represent the gold standard, “ground truth” set of data compared to AI-based tools that rely on images collected from search engines. In a proof-of-concept study published last year, the model was found to be 84 percent accurate in detecting “normal” versus “abnormal” middle ears.
Human versus machine
In the new study, researchers compared the accuracy of a refined model against that of physicians. More than 639 images of the tympanic membrane from children aged 18 years or younger who were undergoing tube placement or surgery to remove fluid from the ear were used to train the model. The images were tagged as “normal,” “infected,” or “fluid behind the eardrum,” as opposed to the “normal” or “abnormal” classification of the team’s previous models. With the segment added, the model achieved an average diagnostic accuracy of 80.8 percent.
A survey was then created asking clinicians and trainees from various medical specialties to view 22 new images of the tympanic membrane and diagnose the ear as one of three tagged categories. While the machine-learning model correctly classified more than 95 percent of the sample images, the average diagnostic score among the 39 physicians who responded to the survey was 65 percent. Additionally, pediatricians and family medicine/general internists correctly classified 60.1 percent and 59.1 percent of the images, respectively.
