EEG Based Functional Brain Network Analysis and Classification of Dyslexic Children During Sustained Attention Task

Seshadri, N. P. Guhan; Singh, Bikesh Kumar; Pachori, Ram Bilas · 2023 · Crossref

DOI: 10.1109/tnsre.2023.3335806

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the functional brain network connectivity of children with developmental dyslexia during a sustained attention task, addressing the gap in literature regarding attention-related neural dynamics in this population. While dyslexia is traditionally linked to phonological processing deficits, evidence suggests attentional impairments also contribute to reading difficulties. The research aims to characterize these network alterations using electroencephalogram (EEG) data and graph theory, and to evaluate the efficacy of machine learning classifiers in distinguishing dyslexic from non-dyslexic children based on these neural features. The study included 30 participants: 15 dyslexic children (mean age 9.83 years) and 15 non-dyslexic controls (mean age 9.91 years), matched for age, sex, and non-verbal IQ. Participants performed a visual continuous performance task (VCPT) requiring sustained attention while 19-channel EEG signals were recorded. Preprocessing involved re-referencing, moving average filtering, and wavelet denoising to remove artifacts. Functional connectivity was estimated using spectral coherence, and brain networks were analyzed using graph theory metrics, including clustering coefficient, characteristic path length, global and local efficiency, and small-worldness. Task-induced changes in these metrics were calculated by comparing task performance to baseline recordings. Two classifiers, Support Vector Machine (SVM) and k-Nearest Neighbor (KNN), were trained on these network features using 5-fold and leave-one-subject-out cross-validation to classify the groups. Results indicated that dyslexic children exhibited significantly higher omission and commission errors during the VCPT, reflecting distractibility and impulsivity. Neurophysiologically, the dyslexic group displayed disrupted functional network properties, particularly in the theta and alpha frequency bands. Specifically, dyslexic children showed a lower clustering coefficient, longer characteristic path length, and reduced global and local efficiency compared to controls, indicating poor functional segregation and impaired information transfer. Notably, the dyslexic group lacked the small-world network organization typically observed in healthy brains. In terms of classification, the KNN classifier achieved the highest performance, reaching a maximum accuracy of 96.7% in distinguishing dyslexic from non-dyslexic subjects. The findings suggest that dyslexia is associated with distinct alterations in functional brain network connectivity during sustained attention, characterized by inefficient information processing and reduced network integration. The high classification accuracy demonstrates the potential of EEG-based graph theoretical features combined with machine learning as objective biomarkers for early and accurate identification of dyslexia, offering a promising alternative to traditional behavioral assessments.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• neuro workload indices