Automated-EEG-Sleep-Staging

A MATLAB toolbox for EEG-based Sleep Stage Classification from preprocessing, feature extraction, feature selection, dimension reduction, and classification using SVM and KNN.

Codes and data for the following paper are extended to different methods:

Diffuse to fuse EEG spectra–intrinsic geometry of sleep dynamics for classification.

1. Introduction.

This package includes the prototype MATLAB codes for Automated EEG Sleep Staging.

The implemented methodes include:

1. Various feature extraction methods, including

   • Multiscale permutation entropy
   • Statistical features
   • AR coefficients
   • Spectrul entropy
   • Hjorth parameters mobility and complexity
   • Approximate entropy
   • Lyapunov exponent
   • Correlation dimension
   • Mel-frequency cepstral coefficients

2. Several dimension reduction methods including PCA, LDA and TSNE

3. Multiple classifiers SVM, KNN, NeuralNets

2. Usage & Dependency.

Dependency:

 sleep-edf dataset
 https://github.com/sajjadkarimi91/SLDR-supervised-linear-dimensionality-reduction-toolbox
 Kijoon Lee (2022). Fast Approximate Entropy (https://www.mathworks.com/matlabcentral/fileexchange/32427-fast-approximate-entropy), MATLAB Central File Exchange.
 Valentina Unakafova (2022). Permutation entropy (fast algorithm) (https://www.mathworks.com/matlabcentral/fileexchange/44161-permutation-entropy-fast-algorithm)

Usage:

Run "main_run.m" or "main_binary.m" to analyze the sleep staging.