Neuro-Bloom: Machine Learning-based Diagnosis of Learning Disabilities via Gamified Testing

This paper proposes a gamified platform that diagnoses children with specific learning disabilities, namely dyslexia, dysgraphia, dyscalculia, and Attention Deficit Hyperactivity Disorder (ADHD). Learning disabilities are neurobiological disorders where cognitive and scholastic skills are affected; these are largely undiagnosed as it requires more awareness, availability of resources, and also relies on traditional subjective screening methods such as behavioral questionnaires and clinical observations [5, 11]. This scenario becomes all the more challenging in schools due to the unavailability of accessible unsupervised diagnostic tools. The Neuro-Bloom platform described here represents an interactive game-based solution which allows self-administered screening of children in the age group of 6–17 years without the help of parental or teacher supervision. Proposed machine learning models such as Random Forest, Support Vector Machine(SVM), Gradient Boosting, and custom Convolutional Neural Network (CNN) ensembles examine response time, accuracy, handwriting, and attention-based game-playing parameters for the detection of learning difficulties. Integrated model accuracies of 94% for dyslexia, 85.7% for ADHD, 82.7% for dysgraphia along with 90% sensitivity in case of dyscalculia, demonstrate the strength of prediction. To the best of our knowledge, this is the first comprehensive game-based unsupervised screening-cum-intervention tool that has been developed for multiple learning disabilities in the Indian educational scenario. NeuroBloom combines engaging gameplays with ML analytics, and as such, is able to offer a scalable, objective, and accessible early detection and personalized intervention solution by responding to some critical lacunae present in conventional diagnostic approaches.