Artificial Intelligence Based Automated Estimation of Sleep Stages Using Electrocardiograph Signals: A Perspective

The primary goal of this study is to expound the Artificial Intelligence schemes utilized in developing an automated sleep staging sytem. Sleep stages, broadly classified as REM and Non-REM (Rapid Eye Movement) are recognized during sleep studies. Electrocardiogram signal is one among the multiple signals recorded during a sleep study. An effort to bring out the correlation between Electrocardiogram and sleep stages would facilitate in developing an automated screening system for identifying sleep disorders. This study assimilates such researches and their outcomes conducted during the last two decades. It is also emphasized that due to liberal availability of Electrocardiogram data in hospitals, using it to distinguish sleep stages would aid in developing better healthcare. The prime methods identified from the literature are the statistical classifiers and neural network based classifiers.The reports discussed are typical of single night polysomnographic recordings. The collective results are then compared with manually scored sleep stages. Out of the various methods, Support Vector Machines and Detrended Fluctuation analysis are the popular methods owing to their nature of analyzing non stationary signals.