In this study, we aimed to examine the relationship between EEG-based connectivity measures and cognitive impairment in a large cohort of MS patients. EEG recordings were obtained from 250 persons with MS, in conjunction with cognitive data. Cognitive functioning was assessed with the Neuropsychological Screening Battery for MS. A patient was classified as cognitively impaired when scoring below the 5th percentile of a normal population on two or more tests. Weighted EEG-based connectivity (i.e., imaginary coherence and phase locking values) matrices were obtained by within 4 frequency bands: delta (1-4Hz), theta (4-8Hz) alpha (8-12Hz) and beta (13-30Hz). We used machine learning to assess the predictive value of functional connectivity for cognitive status. We used 2 (edge detection methods) x 4 (frequency bands) types of feature sets to predict cognitive impairment in our MS sample. The machine learning analysis using a random forest classifier showed no relationship between connectivity and cognitive status across the feature sets. Given the large sample size, this suggests that while in rest with eyes closed, the (non-linear) relation between connectivity and cognition is likely small.
Van De Steen, F, Laton, J, Denissen, S, Baijot, J, Dhooghe, M, D'haeseleer, M, Rossi, C, Van Schependom, J & Nagels, G 2022, 'Cognitive impairment and brain network organisation in MS patients', Multiple Sclerosis Journal, vol. 28, no. 2, pp. 8-9. https://doi.org/10.1177/13524585221100731
Van De Steen, F., Laton, J., Denissen, S., Baijot, J., Dhooghe, M., D'haeseleer, M., Rossi, C., Van Schependom, J., & Nagels, G. (2022). Cognitive impairment and brain network organisation in MS patients. Multiple Sclerosis Journal, 28(2), 8-9. https://doi.org/10.1177/13524585221100731
@article{68bcbebb8a9c4f8e8fe8473fe2811726,
title = "Cognitive impairment and brain network organisation in MS patients",
abstract = "In this study, we aimed to examine the relationship between EEG-based connectivity measures and cognitive impairment in a large cohort of MS patients. EEG recordings were obtained from 250 persons with MS, in conjunction with cognitive data. Cognitive functioning was assessed with the Neuropsychological Screening Battery for MS. A patient was classified as cognitively impaired when scoring below the 5th percentile of a normal population on two or more tests. Weighted EEG-based connectivity (i.e., imaginary coherence and phase locking values) matrices were obtained by within 4 frequency bands: delta (1-4Hz), theta (4-8Hz) alpha (8-12Hz) and beta (13-30Hz). We used machine learning to assess the predictive value of functional connectivity for cognitive status. We used 2 (edge detection methods) x 4 (frequency bands) types of feature sets to predict cognitive impairment in our MS sample. The machine learning analysis using a random forest classifier showed no relationship between connectivity and cognitive status across the feature sets. Given the large sample size, this suggests that while in rest with eyes closed, the (non-linear) relation between connectivity and cognition is likely small.",
keywords = "adult, classifier, cognition, cognitive defect, cohort analysis, conference abstract, edge detection, electroencephalogram, female, functional connectivity, human, human tissue, machine learning, major clinical study, male, mentally disabled person, nerve cell network, predictive value, random forest, sample size",
author = "{Van De Steen}, F. and J. Laton and S. Denissen and J. Baijot and M. Dhooghe and M. D'haeseleer and C. Rossi and {Van Schependom}, J. and G. Nagels",
year = "2022",
month = jun,
day = "1",
doi = "10.1177/13524585221100731",
language = "English",
volume = "28",
pages = "8--9",
journal = "Multiple Sclerosis Journal",
issn = "1352-4585",
publisher = "SAGE Publications Ltd",
number = "2",
}