Design and Development of Efficient Ensemble Expert System for Obstructive Sleep Apnea Detection System using EEG Signal and Machine Learning Analytics

Abstract

Sleep is a biological process essential for human survival, playing a crucial role in cognitive function, physical restoration, and overall health maintenance. Sleep is often disrupted by various sleep disorders, which affect millions of people worldwide and significantly impact their quality of life. Among these disorders, Obstructive Sleep Apnea (OSA) is the most prevalent one, affecting millions of people worldwide. It is characterized by the recurrent collapse of the upper airway, leading to partial or complete obstruction of air to the lungs. For diagnosing OSA, the standard method is polysomnography, which is a comprehensive and expensive diagnostic procedure requiring specialized personnel and overnight monitoring in a sleep laboratory. Due to these limitations, there is a need for more accessible, cost-effective, and simpler diagnostic solutions. In response to these challenges, several researchers throughout the world have been trying to develop Computer-Aided Diagnostic Systems (CADS) using machine learning algorithms that can be a transformative alternative to traditional diagnostic methods for OSA detection. These systems utilize advanced signal processing techniques and state-of-the-art machine learning models to analyze sleep data efficiently. However, despite these advantages, the existing CADS still has some limitations. Most of these models rely on machine learning models that often face the problem of overfitting and poor generalization. The reliance on limited-scale datasets and limited preprocessing used in past studies may also affect the generalizability of results. Additionally, most of the existing studies have segmented whole-night sleep data into 30-second epochs that may cause the loss of intricate dynamics and interrelationships among various sleep stages and instances of apnea, posing a risk of information loss.