A Survey on EEG Phase Amplitude Coupling to Speech Rhythm for the Prediction of Dyslexia

Developmental dyslexia (DLX) hinders the reading learning process of 5%–12% of the world’s population. Those affected by DLX show difficulties in oral phonological tasks, the biological underpinnings of which are still hotly debated. Current research has shown abnormal brain oscillatory coupling to speech rhythms, a procedure known as ‘entrainment’, key to encode phonological representations of speech units. Therefore, brain entrainment to speech rhythms could be used as features in an automatic diagnostic system. This work explores the use of Phase amplitude coupling (PAC) measures to quantify the entrainment between auditory rhythmic stimuli and Electroencephalography (EEG) signals. PAC features are used to train an interpretable machine learning system for predicting DLX in children, achieving accuracy over 90% for the entrainment between the 40 Hz stimulus and the Gamma band using Heights Ratio PAC. Analysis of the classification model reveal differences in the entrainment at regions typically associated to language, paving the way for an accurate and interpretable DLX diagnosis methodology.