By Marilynn Larkin

April 22, 2022

NEW YORK (Reuters Health) – Smartphone-recorded sleep breathing sounds were used to create “reasonably accurate” obstructive sleep apnea (OSA) prediction models in a cross-sectional study, researchers say, but they are not yet ready for the clinic.

As reported in JAMA Otolaryngology-Head & Neck Surgery, Dr. Jeong-Whun Kim of Seoul National University College of Medicine and colleagues conducted a cross-sectional study among patients (mean age, 48; about 84%, men) who visited the university’s hospital sleep center for snoring or sleep apnea from 2015 to 2019.

Audio recordings during sleep were performed using a smartphone during routine, full-night, in-laboratory polysomnography.

The team used an algorithm to create three different threshold criteria according to an apnea hypopnea index (AHI) threshold of 5, 15, or 30 events per hour.

They then created four statistical models according to noise reduction and sound feature selection from the input sound to predict actual AHI: (1) noise reduction without feature selection; (2) noise reduction with feature selection; (3) neither noise reduction nor feature selection; and (4) feature selection without noise reduction. Feature selection involved selecting relevant features from 508 sound features identified during model creation.

Data were split into training (60.5%) and test (39.5%) sets.

For AHI thresholds of 5, 15, and 30 events/h, accuracies were 88.2%, 82.3%, and 81.7%, and the areas under curve were 0.90, 0.89, and 0.90, respectively.

Further analyses showed that using recorded sounds that had not been denoised, and had only selected attributes achieved the highest correlation coefficient.

The authors conclude, “Future research should focus on real-life recordings using various smartphone devices.”

Dr. Temitayo Oyegbile-Chidi of UC Davis Health, in Sacramento, California, a board member of The National Sleep Foundation, commented on the study in an email to Reuters Health. “We can get good information from smartphones and sleep apnea apps, but not in the sense of a diagnosis. An OSA diagnosis needs to always be performed by a board-certified clinician.”

“If an individual has been formally diagnosed with sleep apnea with an in-lab or home sleep study with a sleep physician, smartphone-recorded sounds may be helpful specifically to note any changes that may suggest a need for a re-evaluation,” she said.

“Unless a phone is able to connect to a patient’s oxygen levels while asleep, I’d say that phones can provide only a limited amount of information,” Dr. Oyegbile-Chidi noted. “At a patient level, smartphones can show if the user is tossing and turning or snoring a lot…at best, this data can help determine if sleep is a bit more restless.”

“Perhaps in the future, this could be a promising method for diagnosis, but more development is needed,” she added. “Smartphones are wonderful tools for patients to monitor certain elements that can influence their sleep, but not as a way to predict OSA or provide accurate diagnosis.”

SOURCE: https://bit.ly/37trACJ JAMA Otolaryngology-Head & Neck Surgery, online April 14, 2022.