Publication Details
A neurophysiological model based on resting state EEG functional connectivity features for assessing semantic long-term memory performance
Ahmed Amr ()
Yusoff Zuki Mohd ()
Mohamad Saad Mohamad Naufal (UTP)
Malik Aamir Saeed, Ph.D. (DCSY FIT BUT)
Machine Learning Model, EEG, Electroencephalogram, Semantic, Long term memory, functional connectivity
Existing methods for assessing long-term memory (LTM) rely predominantly on psychometric tests or clinical expert observations. In this study, we propose an objective method for evaluating semantic LTM ability using resting-state electroencephalography (EEG) functional connectivity. Data from 68 participants were analysed, deriving functional connectivity from the phase information of EEG theta (4-8 Hz), alpha (8-13 Hz) and gamma (30-45 Hz) frequency bands across the entire scalp at resting state. Participants' responses were recorded during a memory recall task over four sessions. Multiple linear regression was used to model the LTM score. The proposed method successfully predicted LTM retention after 30 min, with performance metrics of F(18,49) = 2.216, p = 0.014, R=0.670; 2 months retention, F(18,45) = 3.057, p < 0.001, R=0.742; 4 months retention, F(18,42) = 2.237, p = 0.016, R=0.700; and 6 months retention, F(18,36) = 1.988, p = 0.039, R=0.706, respectively. Additionally, this method achieved at least 27 points lower in the Bayesian Information Criterion (BIC) compared to the standard psychometric RAPM test across all retention periods. These findings suggest that the semantic LTM ability of healthy young individuals can be objectively quantified using resting-state EEG functional connectivity. This approach holds promise for future applications in understanding and addressing below standard performance in students learning.
@ARTICLE{FITPUB13256,
author = "Hafeez Ullah Amin and Amr Ahmed and Mohd Zuki Yusoff and Naufal Mohamad Saad Mohamad and Saeed Aamir Malik",
title = "A neurophysiological model based on resting state EEG functional connectivity features for assessing semantic long-term memory performance",
pages = "1--9",
journal = "Biomedical Signal Processing and Control",
volume = 99,
number = 1,
year = 2024,
ISSN = "1746-8108",
doi = "10.1016/j.bspc.2024.106799",
language = "english",
url = "https://www.fit.vut.cz/research/publication/13256"
}