Time changes of involvement indexes in a cohort of epileptic patients during a working memory task

Abstract

Working memory (WM) is the mental ability to encode, keep, and manipulate information over a short period of time. It can be noninvasively investigated by electroencephalography (EEG) and implies a certain level of mental involvement that can be evaluated by computing involvement indexes based on the spectral power of EEG rhythms. WM impairment is known to be present in epileptic patients. Thus, this study aims to evaluate if (and how) EEG-derived involvement indexes change over time in a population of epileptic patients during a WM task. The population analyzed comes from a public dataset, acquired and organized by Boran et al. EEG data were recorded while patients were performing a verbal WM task, which included 50 trials of about 8 s each. The EEG signals were analyzed for the preprocessing and extraction of normalized involvement indexes according to the definition presented in the review by Marcantoni et al. (2023). Starting from the spectral power energy of the EEG-derived rhythms, 37 involvement indexes were computed recursively along the EEG for each patient. Results showed that involvement indexes change throughout the WM task, according to the cognitive engagement elicited in its phases. Changes of involvement indexes with respect to the resting cognitive status of patients were most heightened in the frontal region. Moreover, most indexes increase during the task and the indexes that showed a decrease with respect to resting cognitive status are those having high-frequency rhythms (α, β, γ) at the denominator in their definition. In the population here considered, only the patients affected by brain contusion and hippocampal sclerosis seem to deviate from the physiological paradigm. This study has to be intended without the purpose of generalizing the obtained outcomes, which are preliminary and may be confirmed in further studies on more homogeneous and larger epileptic populations.