A New Framework for Automatic Detection of Patients With Mild Cognitive Impairment Using Resting-State EEG Signals.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Mild cognitive impairment (MCI) is an early indicator of Alzheimer's disease (AD), a significant global health concern.
• Early detection of MCI is crucial for identifying individuals at risk of AD and dementia.
• Electroencephalography (EEG) is a primary tool for investigating MCI biomarkers.

Purpose of the Study:

• To develop an automated framework for distinguishing MCI patients from healthy controls using EEG data.
• To introduce and evaluate novel methods for EEG data processing and feature extraction for MCI detection.

Main Methods:

• A framework incorporating noise removal, segmentation, Piecewise Aggregate Approximation (PAA) for data compression, and feature extraction using Permutation Entropy (PE) and Auto-Regressive (AR) models.
• Classification using Extreme Learning Machine (ELM), Support Vector Machine (SVM), and K-Nearest Neighbours (KNN) models.
• Evaluation using a 10-fold cross-validation on a public MCI EEG database.

Main Results:

• The ELM-based model achieved the highest classification accuracy of 98.78%.
• The proposed ELM method demonstrated a low execution time of 0.281 seconds.
• The framework outperformed existing methods in detecting MCI from EEG data.

Conclusions:

• The developed automated framework provides a robust and efficient biomarker for MCI detection.
• The findings suggest the potential of EEG-based machine learning models for early diagnosis of neurodegenerative diseases like AD.
• The ELM model shows significant promise for clinical application in identifying MCI patients.