Sleep Apnea Detection through HRV and SPO2 Analysis of Single-Lead ECG Signals Using Machine Learning Algorithms

Sleep apnea is a widespread illness indicated by frequent interruptions in breathing while sleeping, leading to many adverse health effects.Thus, we are motivated to help detect sleep apnea by using modern computing technologies.This paper presents a comprehensive study of non-invasive monitoring and diagnostic approaches for sleep apnea detection through heart rate variability (HRV) analysis of single-lead electrocardiogram (ECG) signals using machine learning models.This study investigated the application of machine learning algorithms like Support Vector Machine (SVM), Random Forest (RF), and Long Short-Term Memory (LSTM) on the Apnea-ECG Database v1.0.0.by PhysioNet.The dataset comprises 34,428 1-minute intervals, with a significant imbalance between normal and apnea episodes.A 5-fold stratified cross-validation method was employed for reliable and accurate analysis.Random Forest achieves the highest accuracy of 91.76%, followed by SVM with 86.66%, and LSTM with 84.57%.Precision, Recall, and F1-score demonstrate the higher accuracy of the random forest algorithm in accurately identifying occurrences of apnea while minimizing the presence of false positives and false negatives.This study underlines the prospects of machine learning in enhancing sleep apnea detection and research, contributing to improved patient outcomes and healthcare management.