Remediating Indian school going children having specific learning disorder with dyslexia using artificial intelligence based structured intervention program

Objectives To develop and evaluate an artificial intelligence (AI) based structured intervention program for Indian school going children with dyslexia

Methods A pre-post intervention study was conducted from January-2020 to January-2022 among school going children (8-16 years) having dyslexia. AIIMS (All India Institute of Medical Sciences) dyslexia remedial program was developed in 2018. It was adapted to enhance learning by AI (web-application based phonological training). Assessment was done using GLAD (Grade level Assessment Device, English), CTOPP-2 (Comprehensive Test of Phonological Processing 2ndedition) and WRAT-5 (Wide Range achievement test, Indian version) scoring at baseline and post intervention at 6 weeks. Functional MRI (task based and resting state) was also done in six children.

Results Twenty-four children underwent AI based intervention for 6 weeks. Mean age was 12.61 ± 2.0 years. A statistically significant improvement was observed in reading abilities (p value <0.001) as assessed by GLAD (English), CTOPP2 composite performance (phonological awareness, phonological memory, alternate phonological awareness domains) and WRAT-5 standard scores (word reading, spelling subtests). Among the core subsets of phonological awareness, children showed maximum improvement in blending words (87.5%) followed by elision (83.3%). Functional MRI revealed an increased resting state connectivity in left hemispheric language networks and brain activation profile in areas concerned with phonological processing during visual reading tasks (words, pseudowords) post-intervention.

Conclusions AI based intervention program for dyslexic children can be effective as reflected by change in assessment scores and activated language networks post-intervention. However future studies with larger sample size and longer intervention time is needed.
Keywords: Dyslexia, Artificial intelligence, Functional MRI, CTOPP2, WRAT5