A Review on Machine Learning for EEG Signal Processing in Bioengineering.

Area of Science:

• Bioengineering
• Computational Neuroscience
• Medical Informatics

Background:

• Electroencephalography (EEG) is a crucial diagnostic tool for various health conditions.
• EEG analysis employs numerous classification methods, with machine learning (ML) being prominent.
• A gap exists in comprehensively reviewing ML applications specifically for EEG in bioengineering.

Purpose of the Study:

• To review and analyze machine learning methods used for EEG analysis in bioengineering applications.
• To assess the effectiveness and characteristics of different ML classifiers for EEG data.
• To provide a comprehensive overview of ML in EEG analysis from 1988 to 2018.

Main Methods:

• Systematic literature review of studies published between 1988 and 2018.
• Analysis of machine learning algorithms applied to electroencephalography data.
• Categorization and comparison of supervised and unsupervised learning methods.

Main Results:

• All primary machine learning methods, including Naive-Bayes, Decision Trees/Random Forests, and Support Vector Machines (SVM), have been applied to EEG classification.
• Supervised learning methods, such as SVM and K-Nearest Neighbors (KNN), demonstrate higher average accuracy compared to unsupervised methods.
• Individual ML methods have limitations, but combined approaches show potential for improved classification accuracy.

Conclusions:

• Machine learning offers a versatile toolkit for EEG analysis in bioengineering.
• Supervised learning approaches are generally more accurate for EEG classification.
• Future research should explore optimized combinations of ML methods to enhance EEG analysis effectiveness.