Related Experiment Video
Updated: Jul 1, 2025
11:01
Examining the Characteristics of Episodic Memory using Event-related Potentials in Patients with Alzheimer's Disease
Published on: August 30, 2011
13.6K
Electroencephalography-based classification of Alzheimer's disease spectrum during computer-based cognitive testing
Seul-Kee Kim1,2, Hayom Kim3, Sang Hee Kim2
1Bionics Research Center, Korea Institute of Science and Technology, Seoul, Republic of Korea.
Scientific Reports
|March 4, 2024
Summary
Early Alzheimer's disease (AD) detection is improved using a new computer task and electroencephalography (EEG). This method accurately identifies mild cognitive impairment (MCI) stages, including non-amnestic MCI (naMCI), by analyzing brain activity during memory encoding.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Cognitive Science
Background:
- Alzheimer's disease (AD) diagnosis relies on early detection of mild cognitive impairment (MCI), but non-amnestic MCI (naMCI) is often overlooked.
- Subjective cognitive decline (SCD) is a critical early indicator of AD spectrum pathology, bridging normal aging and MCI.
- Traditional neuropsychological tests for cognitive assessment can be time-consuming and prone to bias.
Purpose of the Study:
- To develop and validate a computer-based cognitive task combined with electroencephalography (EEG) for early AD diagnosis.
- To differentiate between stages of cognitive decline: SCD, naMCI, amnestic MCI (aMCI), and AD.
- To assess the efficacy of EEG during resting and memory encoding states for multi-class classification of AD spectrum disorders.
Main Methods:
- EEG data were collected from 58 participants (SCD, naMCI, aMCI, AD) during resting and memory encoding states.
- Features reflecting phase, spectral, and temporal brain activity were extracted from EEG signals.
- Nine machine learning and three deep learning models were compared using a Leave-one-subject-out cross-validation strategy for classification.
Main Results:
- The computer-based cognitive task showed significant correlations with existing neurophysiological test scores across all cognitive domains.
- Classification performance was significantly higher during memory encoding states (up to 93.10% accuracy) compared to resting states (up to 67.24% accuracy).
- The cKNN model achieved the highest accuracy for the four-class classification task using memory encoding EEG data.
Conclusions:
- EEG analysis during memory encoding, coupled with a computer-based cognitive task, offers a highly accurate method for early AD diagnosis and staging.
- This approach provides a time-efficient and less biased alternative to traditional cognitive assessments.
- The findings highlight the potential of analyzing cognitive process-related brain waves for improved detection of AD spectrum pathologies.
