Introducing the Hybrid "K-means, RLS" Learning for the RBF Network in Obstructive Apnea Disease Detection using...

Area of Science:

• Biomedical Engineering
• Medical Informatics
• Signal Processing

Background:

• Obstructive sleep apnea (OSA) is a serious, potentially fatal condition.
• Timely diagnosis of OSA is essential for effective prevention and treatment.
• Current diagnostic methods may have limitations in accuracy or computational efficiency.

Purpose of the Study:

• To develop a precise and computationally efficient method for early detection of obstructive sleep apnea (OSA).
• To leverage advanced signal processing techniques for improved OSA diagnosis from ECG signals.

Main Methods:

• Feature selection based on Dual Tree Complex Wavelet (DT-CWT) coefficients from ECG signals.
• Feature extraction using frequency and time domain techniques.
• Application of Spectral Regression Discriminant Analysis (SRDA) for feature selection.
• Classification using a novel hybrid Radial Basis Function (RBF) neural network.

Main Results:

• The proposed method demonstrated a 3% improvement in detection accuracy for OSA.
• A significant reduction of at least 30% in computational complexity was achieved compared to recent methods.
• The hybrid RBF network proved less computationally demanding than traditional SVM networks.

Conclusions:

• The developed method offers a precise and computationally efficient approach for early OSA detection.
• Utilizing DT-CWT coefficients with SRDA and a hybrid RBF network shows promise for improving OSA diagnosis.
• This approach could lead to more accessible and effective screening for obstructive sleep apnea.