Sleep disorder and apnea events detection framework with high performance using two-tier learning model design.

Area of Science:

• Medical Technology
• Artificial Intelligence in Medicine
• Sleep Medicine

Background:

• Sleep apnea is a prevalent breathing disorder impacting sleep quality and overall health.
• Manual diagnosis of sleep apnea via polysomnography (PSG) is time-consuming and subjective.
• Accurate and efficient detection methods are crucial for timely patient intervention.

Purpose of the Study:

• To develop and evaluate a novel two-layer deep learning model for automated sleep apnea detection.
• To improve the accuracy and efficiency of diagnosing sleep apnea and its subtypes.
• To provide a robust, patient-independent system for sleep clinics.

Main Methods:

• A two-layer architecture combining deep learning models (DNN, GRU, RNN, LSTM) in the first layer and a machine learning meta-learner in the second.
• Training utilized a 23-feature vector including snore, oxygen saturation, arousal, and sleep scores, alongside PSG data.
• Data pre-processing and under-sampling techniques were applied to address class imbalance in a dataset of 50 pediatric and adult patients.

Main Results:

• The proposed method achieved 95.74% accuracy in detecting apnea, hypopnea, and normal breathing.
• It demonstrated 99.4% accuracy in classifying apnea types (central, mixed, obstructive).
• Experimental results confirmed high accuracy and patient-independent consistency.

Conclusions:

• The developed two-layer deep learning model offers a highly accurate and robust solution for sleep apnea detection.
• This system can significantly assist sleep clinics in detailed sleep disorder diagnosis.
• The model's performance suggests potential for widespread clinical adoption to improve patient outcomes.