disclosures: Cho reports that he has received a licensed domestic patent related to this study. See the study for the relevant financial disclosures from all other authors.
Sleep breathing sounds recorded via smartphone provided fair prediction of obstructive sleep apnea, researchers reported in JAMA Otolaryngology † head † neck surgery†
“The gold standard diagnostic method for OSA is full-night lab polysomnography recording numerous physiological signals that are manually scored by certified sleep technicians or physicians. Therefore, lab polysomnography is expensive and accessibility to a sleep facility is not always easy.” Sung-Woo Cho, MD, of the department of ENT – head and neck surgery at Seoul National University Bundang Hospital at Seoul National University College of Medicine in Seongnam, South Korea, and colleagues wrote. “Given the high prevalence of OSA, it may not be practical for all patients to perform polysomnography in the lab all night.”
The cross-sectional study enrolled 423 patients (mean age 48.1 years; 84.1% males) who visited the sleep center of National University Bundang Hospital in Seoul from August 2015 to August 2019 for snoring or sleep apnea. Patients recorded audio with a smartphone during sleep during routine, nocturnal laboratory polysomnography. Researchers performed binary classifications for various threshold criteria based on an apnea-hyppnea index threshold of five, 15, or 30 events per hour and created four regression models, including noise reduction without feature selection, noise reduction with feature selection, neither noise reduction, nor feature selection, and feature selection without noise reduction. .
Researchers split the data into training (n = 256) and test (n = 167) data sets and patients were grouped as normal (n = 43), mild OSA (n = 80), moderate OSA (n = 109), or severe OSA (n = 191).
Audio recorded with smartphones yielded an accuracy of 88.2% for an apnea-hypopnea index threshold of five events per hour, 82.3% for 15 events per hour, and 81.7% for 30 events per hour. Areas under the curve were 0.9, 0.89 and 0.9 for five, 15 and 30 events per hour, respectively.
All four regression models showed similar results with correlation coefficients ranging from 0.77 to 0.78. Audio recorded on smartphones that was not noiseless and had only selected attributes yielded the highest correlation coefficient in the regression analysis (r = 0.78). Both the apnea-hypopnea index (beta = 0.33) and sleep efficiency (beta = -.2) were associated with an OSA estimation error.
“These prediction models gave reasonable prediction performance and we found that noise reduction was not required for good prediction performance. … Future research should be expanded to include real-life smartphone recordings at home using various smartphone devices,” the researchers wrote. .