Related Experiment Video
Updated: May 6, 2026
Drug-Induced Sleep Endoscopy DISE with Target Controlled Infusion TCI and Bispectral Analysis in Obstructive Sleep Apnea
Published on: December 6, 2016
Pediatric obstructive sleep apnea diagnosis: leveraging machine learning with linear discriminant analysis
Han Qin1, Liping Zhang2, Xiaodan Li3
1Department of Child Health Care, Children's Hospital Capital Institute of Pediatrics, Chinese Academy of Medical Sciences & Peking Union Medical College, Capital Institute of Pediatrics, Beijing, China.
Insights
A machine learning model effectively diagnoses obstructive sleep apnea (OSA) in children using clinical data, offering a feasible alternative to polysomnography (PSG) for screening.
Area of Science:
- Pediatric Sleep Medicine
- Artificial Intelligence in Healthcare
- Clinical Diagnostics
Background:
- Obstructive sleep apnea (OSA) diagnosis in children often relies on polysomnography (PSG), a resource-intensive nocturnal sleep study.
- Clinical features obtainable outside of sleep labs are underutilized for OSA screening in pediatric populations.
- Developing non-invasive, accessible diagnostic tools for pediatric OSA is a significant clinical need.
Purpose of the Study:
- To evaluate the efficacy of a machine learning (ML) algorithm for diagnosing pediatric obstructive sleep apnea (OSA).
- To identify key clinical features for OSA diagnosis in children that can be collected in non-medical settings.
- To compare the performance of an ML-based diagnostic model against traditional PSG questionnaire-based models.
Main Methods:
- A cohort of 2464 children (aged 3-18) suspected of OSA underwent clinical data collection and PSG.
- Data were randomly split into training (80%) and testing (20%) sets.
- Elastic Net algorithm was employed for feature selection, followed by model development and validation using stratified 10-fold cross-validation.
Main Results:
- Machine learning models utilizing Elastic Net-selected features achieved an Area Under the Curve (AUC) of 0.73 for AHI ≥5 and 0.78 for AHI ≥10.
- The selected features outperformed those derived from PSG questionnaires.
- Linear Discriminant Analysis demonstrated 44% sensitivity and 90% specificity for OSA identification, indicating clinical utility.
Conclusions:
- A machine learning model leveraging children's clinical features demonstrates effectiveness in diagnosing OSA.
- This approach offers a practical, non-nocturnal alternative for OSA screening and severity stratification in pediatric patients.
- The developed ML model presents a promising tool to improve accessibility and efficiency in pediatric OSA diagnosis.
Objective:
The objective of this study was to investigate the effectiveness of a machine learning algorithm in diagnosing OSA in children based on clinical features that can be obtained in nonnocturnal and nonmedical environments.
Patients And Methods:
This study was conducted at Beijing Children's Hospital from April 2018 to October 2019. The participants in this study were 2464 children aged 3-18 suspected of having OSA who underwent clinical data collection and polysomnography(PSG). Participants' data were randomly divided into a training set and a testing set at a ratio of 8:2. The elastic net algorithm was used for feature selection to simplify the model. Stratified 10-fold cross-validation was repeated five times to ensure the robustness of the results.
Results:
Feature selection using Elastic Net resulted in 47 features for AHI ≥5 and 31 features for AHI ≥10 being retained. The machine learning model using these selected features achieved an average AUC of 0.73 for AHI ≥5 and 0.78 for AHI ≥10 when tested externally, outperforming models based on PSG questionnaire features. Linear Discriminant Analysis using the selected features identified OSA with a sensitivity of 44% and specificity of 90%, providing a feasible clinical alternative to PSG for stratifying OSA severity.
Conclusions:
This study shows that a machine learning model based on children's clinical features effectively identifies OSA in children. Establishing a machine learning screening model based on the clinical features of the target population may be a feasible clinical alternative to nocturnal OSA sleep diagnosis.
