University of Pittsburgh researchers have developed a smartphone app that can use artificial intelligence to diagnose ear infections or acute otitis media with over 90% accuracy. The development will help prevent incorrect antibiotic treatment in children due to false diagnosis of diseases.
The success rate of diagnosing acute otitis media among medical staff can vary from 30% to 84%. Acute otitis media is often misdiagnosed, confused with similar diseases, and as a result, patients are prescribed unnecessary antibiotics.
A new application developed by scientists from Pittsburgh will help you quickly make the correct diagnosis and choose treatment. The software analyzes video of the eardrum captured using an otoscope connected to a smartphone. Next, the software studies the condition of the membrane and issues a diagnosis.
To develop the application, the neural network analyzed 921 video recordings of eardrums in 635 children observed in pediatric departments over the past five years. Another 230 videos were used to test the technology. The accuracy of correct diagnosis was increased to 93%. Researchers hope that the program will help make the correct diagnosis and prescribe treatment faster.