Smartphone-Based Detection of Middle Ear Fluid,
Jennifer Abbasi
JAMA. 2019;322(2):107. doi:10.1001/jama.2019.9395
Researchers are developing a way for both physicians and parents to detect and monitor middle ear fluid in children with nothing more than a smartphone and a piece of paper. In a proof-of-concept study published in Science Translational Medicine, clinicians used the new smartphone-based system to detect middle ear fluid in 98 pediatric ears with 85% sensitivity and 82% specificity, accuracy comparable with pneumatic otoscopy and tympanometry, the recommended techniques. Parents using the system classified 25 out of 26 ears as accurately as clinicians using the system.

Detecting middle ear fluid is essential for diagnosing children’s ear conditions, including acute ear infections, according to researcher Sharat Raju, MD, of the University of Washington in Seattle. Visual assessment is the standard diagnostic method, but it’s often unreliable. Pneumatic otoscopy and tympanometry are more accurate but are either technically difficult or require specialized equipment.

With the new system, users download a smartphone application, through which they play a chirp of sound into the child’s ear canal. The sound reflects off the ear drum and is picked up by the phone’s microphone. A machine learning algorithm then interprets the sound and determines whether fluid is present. The only attachment needed is a small, easy-to-construct paper funnel.

As with tympanometry and pneumatic otoscopy, the system detects middle ear fluid by evaluating eardrum mobility. Ear problems affecting eardrum mobility could result in false-positives.

The researchers are now refining the algorithm and deploying the system on a wider range of smartphone models, and they plan to test it in a larger, independent data set.