Thanks to artificial intelligence, American researchers have transformed headsets into a medical diagnostic tool. By analyzing the resonance of an audio signal, they can detect ear plugs, ruptured eardrums and ear infections.
L’emergence deep learning, or deep learning over the past ten years has enabled great advances in matter artificial intelligence (AI). Thanks to neural networks, it makes it possible to train algorithms for many tasks, such as facial recognition, machine translation, deep fakes and even to diagnose certain diseases. And indeed, researchers from State University of New York at Buffalo have developed an AI capable of checking the health of the ears.
The system was named Ear Health (ear health) and works through in-ear headphones connected to a smartphone equipped with a platform deep learning. The headphones emit a chirps, or frequency ramp modulated pulse. The AI uses the built-in microphone to measure how this sound signal resonates in the ear canals and thus creates a profile of the unique geometry of the inner ear of the wearer.
The system will then issue chirps on a regular basis, for example once a day, to check that this profile has not changed. It is able to detect three conditions that modify the geometry of the auditory canals: a cerumen cap a rupture of the ear-drum and an otitis.
The researchers tested their system on 92 volunteers, consisting of 27 people without ear problems, 22 people with a ruptured eardrum25 people suffering from otitis and finally 18 people with a plug of cerumen. The AI managed to correctly identify the conditions in 82.6% of cases. The next step will be to study how other factors, such as body hair or previous inflammations eardrum may influence these results.