A Review on Automated Sleep Study.

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Sleep Medicine

Background:

• Automated sleep analysis is rapidly advancing, crucial for understanding sleep's health impacts.
• A systematic review of 87 papers was conducted across major scientific databases.
• Focus was on methods, signal types, and machine learning algorithms in automated sleep analysis.

Purpose of the Study:

• To critically evaluate the strengths and weaknesses of current automated sleep analysis methods.
• To identify the current landscape and future directions in sleep research.
• To provide a foundation for researchers and practitioners in sleep diagnostics.

Main Methods:

• Systematic literature review of 87 papers from Google Scholar, PubMed, IEEE Xplore, and ScienceDirect.
• Prioritization of studies based on methods, signal modalities, and machine learning algorithms.
• Critical evaluation of reported accuracies, strengths, and weaknesses of various approaches.

Main Results:

• K-Nearest Neighbors (KNN), Ensemble Learning, and Support Vector Machine (SVM) show high accuracy but have performance variability and computational demands.
• Integrating traditional feature extraction with deep learning and combining deep neural networks show promise for enhanced diagnostic accuracy.
• Challenges include performance variability and computational costs, requiring careful classifier selection.

Conclusions:

• The field needs adaptive classifiers, cross-modality integration, and collaboration for accurate, robust, and accessible sleep diagnostics.
• Future research should focus on more effective and nuanced approaches to sleep analysis.
• This review synthesizes current knowledge, guiding future advancements in automated sleep diagnostics.