Related Experiment Videos
Integrating Complete Blood Count Parameters with Demographic Characteristics for Obstructive Sleep Apnea Prediction
Jianwei Ge1, Yi Ling1, Yingchen Wang1
1Department of Neurology, First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, 310000, People's Republic of China.
Nature and Science of Sleep
|August 10, 2026
Summary
Machine learning models using complete blood count (CBC) and demographic data effectively predict obstructive sleep apnea (OSA) risk in adults. These accessible biomarkers aid in early risk stratification for suspected OSA patients.
Area of Science:
- Medical Informatics
- Sleep Medicine
- Biomarker Discovery
Background:
- Obstructive sleep apnea (OSA) poses significant health risks.
- Accurate risk stratification is crucial for timely diagnosis and management.
- Integrating readily available clinical data can improve predictive models.
Purpose of the Study:
- To develop and validate machine learning models for OSA risk stratification.
- To integrate complete blood count (CBC) parameters with demographic data.
- To assess model performance in adults with suspected OSA.
Main Methods:
- Retrospective analysis of 5,828 adults with suspected OSA.
- Feature selection using LASSO logistic regression identified 9 predictors.
- Four machine learning algorithms were trained and validated temporally.
- Model performance evaluated using AUC, Brier score, and decision curve analysis.
Main Results:
- Logistic regression model achieved the highest validation performance (AUC, 0.901).
- Key predictors included hemoglobin, BMI, age, gender, and mean corpuscular hemoglobin (MCH).
- Models demonstrated good discriminative ability and calibration across different apnea-hypopnea index (AHI) thresholds.
Conclusions:
- Machine learning models integrating CBC and demographic data effectively stratify OSA risk.
- These accessible biomarkers can aid pre-home sleep apnea testing (HSAT) triage.
- The findings support the use of routine blood tests for OSA risk assessment.