Deviant functional connectivity patterns in the EEG related to developmental dyslexia and their potential use for...

Area of Science:

• Neuroscience
• Developmental Psychology
• Machine Learning Applications

Background:

• Developmental dyslexia (DD) is a common learning disorder with suspected neural underpinnings.
• Electroencephalogram (EEG) measures combined with machine learning show potential for DD screening.
• Previous studies often lack validation on independent datasets, limiting practical application.

Purpose of the Study:

• To develop and validate an EEG-based screening approach for developmental dyslexia in Chinese children.
• To investigate the neural correlates of DD using functional connectivity (FC) analysis.
• To assess the performance of a machine learning model for DD detection using independent samples.

Main Methods:

• EEG data recorded from 130 Chinese children (82 with DD, 48 typically developing) during resting-state and verbal working-memory tasks.
• Functional connectivity (FC) calculated across delta, theta, alpha, and beta bands using Pearson correlation coefficients (PCC), phase-locking value (PLV), and a rho (RHO) measure.
• A convolutional neural network (CNN) was trained and validated on independent participant samples to ensure robustness.

Main Results:

• Beta band connectivity measures, particularly eyes-open RHO and eyes-closed PLV, showed the highest discriminative power.
• The model achieved high within-sample accuracy (97.47%) but a lower, though significant, cross-sample accuracy (64.99%).
• Children with DD exhibited altered temporal-parietal, central, and frontal-central connectivity patterns, correlating with reading abilities.

Conclusions:

• Preliminary evidence suggests functional network abnormalities in Chinese children with DD, identifiable via EEG.
• The developed EEG-based screening approach shows potential but requires further validation in larger, diverse cohorts.
• Current model performance is exploratory and not yet sufficient for clinical deployment without refinement.