Automatic Human Sleep Stage Scoring Using Deep Neural Networks

Frontiers in Neuroscience

Alexander Malafeev¹  Peter Achermann¹  Joachim Buhmann³  Dmitry Laptev³  Stefan Bauer⁴  Ximena Omlin⁶  Aleksandra Wierzbicka⁷  Wojciech Jernajczyk⁷  Adam Wichniak⁸  Robert Riener⁹ 

[1] Center for Interdisciplinary Sleep Research, University of Zurich, Zurich, Switzerland;Chronobiology and Sleep Research, Institute of Pharmacology and Toxicology, University of Zurich, Zurich, Switzerland;Information Science and Engineering, Institute for Machine Learning, ETH Zurich, Zurich, Switzerland;Max Planck Institute for Intelligent Systems, Tübingen, Germany;Neuroscience Center Zurich, University of Zurich and ETH Zurich, Zurich, Switzerland;Sensory-Motor Systems Lab, ETH Zurich, Zurich, Switzerland;Sleep Disorders Center, Department of Clinical Neurophysiology, Institute of Psychiatry and Neurology in Warsaw, Warsaw, Poland;Third Department of Psychiatry and Sleep Disorders Center, Institute of Psychiatry and Neurology in Warsaw, Warsaw, Poland;University Hospital Balgrist (SCI Center), Medical Faculty, University of Zurich, Zurich, Switzerland;

关键词: deep learning;    sleep;    EEG;    automatic scoring;    random forest;    artificial neural networks;   

DOI  :  10.3389/fnins.2018.00781

来源: DOAJ

【 摘 要 】

The classification of sleep stages is the first and an important step in the quantitative analysis of polysomnographic recordings. Sleep stage scoring relies heavily on visual pattern recognition by a human expert and is time consuming and subjective. Thus, there is a need for automatic classification. In this work we developed machine learning algorithms for sleep classification: random forest (RF) classification based on features and artificial neural networks (ANNs) working both with features and raw data. We tested our methods in healthy subjects and in patients. Most algorithms yielded good results comparable to human interrater agreement. Our study revealed that deep neural networks (DNNs) working with raw data performed better than feature-based methods. We also demonstrated that taking the local temporal structure of sleep into account a priori is important. Our results demonstrate the utility of neural network architectures for the classification of sleep.

【 授权许可】

Unknown