Novel Framework for Autism Spectrum Disorder Identification and Tailored Education With Effective Data Mining and Ensemble Learning Techniques

Abstract

Autism Spectrum Disorder (ASD) is a neurological condition that significantly affects cognitive abilities, language comprehension, object recognition, interpersonal skills, and communication capabilities. Its primary origin is genetic, and early detection and intervention can mitigate the need for costly medical procedures and lengthy examinations for individuals affected by ASD. Autism Spectrum Disorder is highly diverse, with each affected child being unique. It is often stated that no two autistic children are alike, meaning that what benefits one child may not be suitable for another. An effective teaching approach may be challenging to determine for a child with autism. Two ASD screening datasets of toddlers are merged in this study. The Synthetic Minority Oversampling Technique (SMOTE) method to balance the dataset, followed by feature selection methods. The research introduces a two-phase system: the first phase employs various machine learning models, including an ensemble of random forest and XGBoost classifiers that 94% accurate in ASD identification. In the second phase, the study focuses on identifying appropriate teaching methods for children with ASD by evaluating their physical, verbal, and behavioural performance. This research aims to provide personalized educational approaches for individuals with ASD, harnessing machine learning to enhance precision in addressing their unique needs.

Keywords

• Autism spectrum disorder (ASD)
• autism students learning
• SMOTE
• multi-model learning
• feature engineering