Machine Learning Approaches with Automated Sleep Staging System based on Two-Layer Heterogeneous Ensemble Learning Stacking Model

Abstract

Sleep is an essential requirement for human health and well-being, but many people face sleep problems. These problems can lead to several neurological and physical disorders and adversely affect the overall quality of life. Artificial intelligence (AI)based methods for automated sleep stage classification is a fundamental approach to evaluating and treating this public health challenge.The main contribution of this research work is to develop an Automated Sleep Staging System based on Two-Layer Heterogeneous Ensemble Learning Stacking Model (ASSS-TL-HELSM) for sleep staging under the American Academy of Sleep Medicine (AASM) sleep scoring rules. The main aim of this model is to enhance sleep staging accuracy, reduce overfitting and handle overdrift. For signal preprocessing, we use two different feature selection techniques, Fisher Score (FS), and ReliefF (ReF). For feature extraction, we obtain a total of 28 features.The proposed model analyzes the sleep behavior of the subject using the seasonal and trend components. Sleep recordings from two different subgroups of Institute of Systems and Robotics University of Coimbra (ISRUC-Sleep) were obtained for our experiments.Compared with recent studies using single-channel electro encephalogram (EEG) signals, our proposed ASSS-TL-HELSM model shows the best sleep staging classification accuracy performance on a five sleep stages classification (SC-5) task. The overall classification accuracy is 97.93%, and 97% for features selected through FS and ReF respectively, with the subgroup-I(SG-I) data; similarly, for the subgroup-III(SG-III) data, the features selected through FS, and ReF show a classification accuracy of 98.16% and 98.78% respectively. The comparisons between the proposed model and the existing model show that the proposed model gives better sleep staging accuracy for the five-sleep state's classification.

Tree-Based Machine Learning Techniques for Automated Human Sleep Stage Classification

“…With that proposed model they obtained an accuracy of 99.34% on SG-I dataset for 5 stage classification. Another high accuracy work done by Satapathy and Loganathan [51].…”

“…Another point of literature gap which can be seen in the Table 1, the number of individuals in approximately half of the databases used in similar studies is less or the same amount than the number of individuals in ours. In studies [50,51], were conducted on 3 groups and the number of individuals in two of them was less than 50. The number of individuals in the two references is around 60, not much different from 50.…”