ResApp Health, the Australian digital health startup that has been working on software for smartphones that can diagnose respiratory diseases from the sound signatures of coughs (US clinical trials underway), has announced a new area of development – a diagnostic test to diagnose obstructive sleep apnea (OSA) using a smartphone – simply by leaving the smartphone next to one’s bed to record sounds of the subject while sleeping – and then the algorithms provide a diagnosis (similar AI use to the respiratory software).
OSA is currently diagnosed using sleep laboratory polysomnography or home sleep testing that utilizes specialized equipment to monitor heart, lung and brain activity, breathing patterns, arm and leg movements, and blood oxygen levels while asleep. Neither method offers the ability to mass screen patients due to high cost and low availability, and many patients find traditional testing to be complicated and uncomfortable.
To resolve this, ResApp has developed new machine-learning algorithms to measure the severity of OSA from a patient’s overnight breathing and snoring sounds recorded using a smartphone placed on a bedside table. The company is working with Dr. Philip Currie and Dr. Ivan Ling of Cardio Respiratory Sleep (CRS) to recruit patients at Hollywood Private Hospital and The Park Private Hospital in Perth, Australia.
Recent data from the Wisconsin Sleep Cohort Study showed that sleep apnea affects more than three in 10 men and nearly two in 10 women. Eighty percent of people suffering moderate and severe sleep apnea are undiagnosed. Untreated OSA is known to increase the risk of heart disease, hypertension, stroke and type 2 diabetes, and is estimated by the American Academy of Sleep Medicine (AASM) to cost the US economy $149.6 billion annually.
Proof of Concept Key Findings
Preliminary results from ResApp’s study achieved 86% sensitivity and 83% specificity for identifying patients with an apnea hypopnea index (AHI) greater than or equal to 15 (patients with moderate and severe sleep apnea) compared with simultaneous in-laboratory polysomnography scored using the current 2012 AASM scoring criteria. The area under the receiver operating characteristic curve (a standard measure of how well a test distinguishes between two diagnostic groups, where a value of 1 represents a perfect test) was 0.91.
Similar results were obtained for identifying patients with AHI greater than or equal to 30 (patients with severe sleep apnoea). These results were obtained using ten-fold cross-validation on a large cohort of 731 patients, 62% of which were male. The mean age of patients was 53 (range 18-87) with a mean AHI of 24 (range 0-196). The company is currently recruiting patients for a prospective study and is targeting a regulatory submission (in Australia) by the end of this calendar year.
“We are very excited about these excellent preliminary results for identifying OSA,” said Tony Keating, CEO and Managing Director of ResApp. “There is a strong clinical and economic need for reducing the number of undiagnosed sleep apnea sufferers and by utilizing a smartphone we have the opportunity to deliver a highly-scalable, accurate and easy to use screening test to the mass market. By leveraging our expertise in using audio signatures to identify respiratory conditions we have created another large commercial opportunity.”