An Ensemble Feature Selection Approach to Identify Relevant Features from EEG Signals

Abstract

Identifying relevant data to support the automatic analysis of electroencephalograms (EEG) has become a challenge. Although there are many proposals to support the diagnosis of neurological pathologies, the current challenge is to improve the reliability of the tools to classify or detect abnormalities. In this study, we used an ensemble feature selection approach to integrate the advantages of several feature selection algorithms to improve the identification of the characteristics with high power of differentiation in the classification of normal and abnormal EEG signals. Discrimination was evaluated using several classifiers, i.e., decision tree, logistic regression, random forest, and Support Vecctor Machine (SVM); furthermore, performance was assessed by accuracy, specificity, and sensitivity metrics. The evaluation results showed that Ensemble Feature Selection (EFS) is a helpful tool to select relevant features from the EEGs. Thus, the stability calculated for the EFS method proposed was almost perfect in most of the cases evaluated. Moreover, the assessed classifiers evidenced that the models improved in performance when trained with the EFS approach’s features. In addition, the classifier of epileptiform events built using the features selected by the EFS method achieved an accuracy, sensitivity, and specificity of 97.64%, 96.78%, and 97.95%, respectively; finally, the stability of the EFS method evidenced a reliable subset of relevant features. Moreover, the accuracy, sensitivity, and specificity of the EEG detector are equal to or greater than the values reported in the literature.

Efficient Generalized EEG-Based Drowsiness Detection Approach with Minimal Electrodes

“…These methods aim to improve model performance by reducing dimensionality, improving interpretability, and mitigating overfitting. Common approaches include filtering methods [46], which evaluate characteristics independently of the learning algorithm; encapsulation methods [47], which use the performance of the learning algorithm as a feature selection criterion; and embedded methods [48], where feature selection is integrated into the model building process itself. Each method offers distinct advantages and trade-offs, depending on factors such as the size of the dataset, dimensionality, and computing resources.…”

An Adaptive Matrix-Based Evolutionary Computation Framework for EEG Feature Selection

“…This approach includes techniques such as decision trees [31], random forests [32], and multinomial logistic regression [33], which inherently evaluate the importance of features during model construction. For example, Maritza et al [34] adopted an embedded feature selection method to improve the identification of relevant features in EEG signals and to enhance the performance of the classifier in detecting anomalies, achieving high accuracy and stability. Albaqami et al [35] proposed a feature dimensionality reduction algorithm and adopted gradient boosting decision tree for feature extraction and classification.…”

Digital Transformation in Epilepsy Diagnosis Using Raw Images and Transfer Learning in Electroencephalograms

“…To perform this work and others related [5,18,[29][30][31], 100 pediatric EEGs were collected from the same number of children aged between 22 days and 17 years old, suspected of suffering epilepsy. The exams were performed with the patients asleep, on the recommendation of the neurologist.…”