Prediction of Obstructive Sleep Apnea Based on Respiratory Sounds Recorded Between Sleep Onset and Sleep Offset
- Author:
Jeong Whun KIM
1
;
Taehoon KIM
;
Jaeyoung SHIN
;
Goun CHOE
;
Hyun Jung LIM
;
Chae Seo RHEE
;
Kyogu LEE
;
Sung Woo CHO
Author Information
- Publication Type:Original Article
- Keywords: Obstructive Sleep Apnea; Respiratory Sounds; Polysomnography; Machine Learning
- MeSH: Apnea; Area Under Curve; Body Mass Index; Classification; Humans; Logistic Models; Machine Learning; Noise; Polysomnography; Respiratory Sounds; ROC Curve; Sensitivity and Specificity; Sleep Apnea, Obstructive; Sleep Stages; Sleep, REM
- From:Clinical and Experimental Otorhinolaryngology 2019;12(1):72-78
- CountryRepublic of Korea
- Language:English
- Abstract: OBJECTIVES: To develop a simple algorithm for prescreening of obstructive sleep apnea (OSA) on the basis of respiratorysounds recorded during polysomnography during all sleep stages between sleep onset and offset. METHODS: Patients who underwent attended, in-laboratory, full-night polysomnography were included. For all patients, audiorecordings were performed with an air-conduction microphone during polysomnography. Analyses included allsleep stages (i.e., N1, N2, N3, rapid eye movement, and waking). After noise reduction preprocessing, data were segmentedinto 5-s windows and sound features were extracted. Prediction models were established and validated with10-fold cross-validation by using simple logistic regression. Binary classifications were separately conducted for threedifferent threshold criteria at apnea hypopnea index (AHI) of 5, 15, or 30. Prediction model characteristics, includingaccuracy, sensitivity, specificity, positive predictive value (precision), negative predictive value, and area under thecurve (AUC) of the receiver operating characteristic were computed. RESULTS: A total of 116 subjects were included; their mean age, body mass index, and AHI were 50.4 years, 25.5 kg/m2, and23.0/hr, respectively. A total of 508 sound features were extracted from respiratory sounds recorded throughoutsleep. Accuracies of binary classifiers at AHIs of 5, 15, and 30 were 82.7%, 84.4%, and 85.3%, respectively. Predictionperformances for the classifiers at AHIs of 5, 15, and 30 were AUC, 0.83, 0.901, and 0.91; sensitivity, 87.5%,81.6%, and 60%; and specificity, 67.8%, 87.5%, and 94.1%. Respective precision values of the classifiers were89.5%, 87.5%, and 78.2% for AHIs of 5, 15, and 30. CONCLUSION: This study showed that our binary classifier predicted patients with AHI of ≥15 with sensitivity and specificityof >80% by using respiratory sounds during sleep. Since our prediction model included all sleep stage data, algorithmsbased on respiratory sounds may have a high value for prescreening OSA with mobile devices.
