| Popis: |
Understanding how global brain networks are affected in mild cognitive impairment may explain the changes concerning the electrophysiology of the brain. We research functional changes within neuronal networks in frontal, temporal, parietal lobe and central zone using graph theory in visual oddball paradigm. Thirty subjects (19 females; mean age of 70.63 years old, age range 61 to 79 years old, 11 males; mean age of 70.36 years old, and age range 63 to 81 years old) 14 with mild cognitive impairment and 16 health control. Oddball protocol with target and no target stimuli electroencephalography (EEG) signal were analyzed through functional connectivity (FC) along with Change point detection method and a detailed preprocessing in order to detect dynamical changes in the signal, which reflect the non-stationary behavior of a given EEG signal. Linear Discriminant Analysis (LDA), K-Nearest Neighbor (KNN) and Decision Tree (DT) were trained to classify brain activity in Mild Cognitive Impairment (MCI) and Health Control (HC) subjects based on graph measurements and the physiology interpretation of the Mild Cognitive Impairment. The findings show that using a combination of EEG-based model-free functional connectivity measurements and machine learning to categorize cognitive impairment is effective. The research can be expanded upon to study the possibilities of identifying cognitive decline in real-time, dynamic, and complicated real-world circumstances. |
