A brain connectivity characterization of children with different levels of mathematical achievement based on graph metrics

Abstract

Recent studies aiming to facilitate mathematical skill development in primary school children have explored the electrophysiological characteristics associated with different levels of arithmetic achievement. The present work introduces an alternative EEG signal characterization using graph metrics and, based on such features, a classification analysis using a decision tree model. This proposal aims to identify group differences in brain connectivity networks with respect to mathematical skills in elementary school children. The methods of analysis utilized were signal-processing (EEG artifact removal, Laplacian filtering, and magnitude square coherence measurement) and the characterization (Graph metrics) and classification (Decision Tree) of EEG signals recorded during performance of a numerical comparison task. Our results suggest that the analysis of quantitative EEG frequency-band parameters can be used successfully to discriminate several levels of arithmetic achievement. Specifically, the most significant results showed an accuracy of 80.00% (α band), 78.33% (δ band), and 76.67% (θ band) in differentiating high-skilled participants from lowskilled ones, averaged-skilled subjects from all others, and averaged-skilled participants from low-skilled ones, respectively. The use of a decision tree tool during the classification stage allows the identification of several brain areas that seem to be more specialized in numerical processing.

Resting State EEG Related to Mathematical Improvement After Spatial Training in Children

“…The correlation between fronto-parietal connectivity and mathematics was also reported for EEG recorded in other mathematical tasks. For example, the fronto-parietal coherence ( González-Garrido et al, 2018 ) and frontal/parietal based coherence networks ( Torres-Ramos et al, 2020 ) in a numerical comparison task are correlated with different levels of mathematical achievement in children, and the fronto-parietal power correlation in a logical-mathematical task is correlated with the behavioral outcome of the task in adults ( Molina del Río et al, 2019 ).…”

From brain to education through machine learning: Predicting literacy and numeracy skills from neuroimaging data

“…Among the studies included in the current review, LOOCV was the most widely adopted (21 studies), while most other studies used 10-fold CV (10 studies). Although a recent study suggested that repetitive cross-validation is more reliable than the leave-one-out method (Valente et al, 2021), this technique was only adopted in three studies (Litwińczuk et al, 2023;Nemmi et al, 2023;Torres-Ramos et al, 2020). Overall, there is wide variability in the machine-learning techniques used in neuroimaging studies, both in terms of algorithm selection and CV method.…”

Dyslexia, the Amsterdam Way

“…Although our findings should be interpreted with caution, they contribute to the rapidly growing literature showing that machine learning classification of dyslexia based on features extracted from the EEG holds considerable promise-in particular, when classification is based on tasks specifically designed to examine those processes that are thought to be compromised in dyslexia. A recent study focusing on mathematical achievement provides an interesting illustration of such an approach [75]. This study included children who scored low, average, or high on the math section of an achievement test and were asked to perform a numerical comparison task while their EEG was recorded.…”