Classification of Mild Cognitive Impairment from multi-domain features of resting-state EEG

In recent years, electroencephalography (EEG) has emerged as a low-cost, accessible and objective tools for the early diagnosis of Alzheimer's disease (AD). AD is preceded by Mild Cognitive Impairment (MCI), typically refers to early-stage AD disease. The purpose of this study is to classify MCI patients from the multi-domain features of their electroencephalography (EEG). Firstly, we extracted the multi-domain (time, frequency and information theory) features from resting-state EEG signals before and after a cognitive task from 15 MCI groups and 15 age-matched healthy controls. Then, principal component analysis (PCA) was used to perform feature selection. After that, we compared the performance between SVM and KNN on our EEG dataset. The good performance was observed both from SVM and KNN, which demonstrates the effectiveness of multi-domain features. Furthermore, KNN performs better than SVM and the EEG signals after the cognitive task works better than those before the task.