When used by trained clinicians, the study found an 85% sensitivity and 82% specificity with the use of this smartphone technology and augmented machine learning system.
“The nice thing about this study is that it was done primarily in surgical patients, specifically in children getting ear tubes,” said Dr. Raju, adding that they tested the smartphone technology in these patients because a definitive diagnosis of fluid behind the tympanic membrane can be made after surgical placement of ear tubes.
When they compared the smartphone technology system with the older medical equipment and systems used for diagnosis of tympanic membrane diseases, the investigators found significantly improved performance values with the smartphone technology using modern machine learning techniques.
High Diagnostic Performance Across Platforms and Practitioners
The study also found that the high sensitivity and specificity found with the smartphone technology were similar across various mobile platforms. The first clinical study included the use of both the iPhone 5s and Samsung Galaxy S6. Other mobile technologies tested in a subset of patients with similar performance outcomes included iPhone 6s and Galaxy S7.
The study also included assessment of performance when used by parents in the clinical setting. After a brief demonstration of how to use the technology by a trained clinician, parents of patients (n = 25 ears) using the technology achieved comparable results to those of the trained clinician.
Dr. Raju emphasized that the main role of this technology as he envisions it is to put it in the hands of providers and parents for whom the older, more costly technologies are not accessible or available. “I think pneumatic otoscopy or tympanometry still have critical roles in a physician office, but in places where providers don’t have access to those techniques or in situations where a patient at home can track fluid over time, this is where this technology comes in handy,” he said.
Going forward, he emphasized the need for additional testing at different sites and suggested also using the system in a fresh cohort of patients.