Dyslexia: automated diagnosis by computerized classification of brain electrical activity.

We describe a method for the diagnosis of dyslexia based upon a study of electroencephalographic and evoked potential data recorded from 13 normal and 11 dyslexic boys. Measurements were made from topographic maps of brain electrical activity recorded during resting and activated testing conditions. Using a statistically based technique, we developed rules for classification that successfully diagnosed 80 to 90% of subjects not used in the initial rule development. The nature of the most useful measurements suggests that aberrant neurophysiology in dyslexia involves both hemispheres and is present at rest as well as during complex testing. The method has promise for future diagnosis and research.