Sleep apnea classification based on respiration signals by using ensemble methods

Area of Science:

• Biomedical Engineering
• Signal Processing
• Sleep Medicine

Background:

• Sleep apnea is a common disorder with significant health implications.
• Accurate and efficient detection methods are crucial for timely diagnosis and treatment.
• Current detection methods can be complex and time-consuming.

Purpose of the Study:

• To develop an efficient and robust method for classifying minute-based sleep apnea occurrences.
• To evaluate the performance of ensemble classifiers using wavelet transform and principal component analysis on respiration signals.

Main Methods:

• Respiration signals from abdominal, chest, and nasal routes were extracted from polysomnography recordings.
• Wavelet transform was used for feature extraction on 1-minute signal segments.
• Principal component analysis facilitated dimension reduction.
• Eight recordings were used to train and test three ensemble classifiers: AdaBoost, Random Forest, and Random Subspace.

Main Results:

• Classification accuracies ranged from 92.07% to 98.43% for abdominal signals.
• Accuracies ranged from 92.75% to 98.68% for chest signals.
• Accuracies ranged from 92.42% to 98.61% for nasal signals.
• The highest accuracy of 98.68% was achieved using the Random Forest classifier on nasal signals.

Conclusions:

• Ensemble classifiers, particularly Random Forest, demonstrate high efficacy in classifying sleep apnea.
• Nasal respiration signals provide a robust basis for accurate sleep apnea detection.
• The developed method offers an efficient and accurate approach for sleep apnea diagnosis.