Enhancing Autism Detection Through Visual Pattern Analysis Based On Machine Learning

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that may be discovered at a young age by analyzing the visual patterns of children's faces, which, unfortunately, requires the expertise of professionals. SVM is a powerful machine-learning technique that can be used in classification tasks, including autism identification. Our study investigates machine learning methods based on Support Vector Machine (SVM) techniques and the effectiveness of using the green channel for early identification of autism through facial visual patterns in children to help automatically diagnose ASD. Our study employs a different feature extraction approach based on the Gray-Level Co-occurrence Matrix to lower the dimensionality of features. Finally, we evaluated different SVM kernels, such as linear, radial basis function (RBF), and polynomial kernels, to evaluate system effectiveness in identifying visual patterns indicative of ASD. The RBF kernel produces the best performance of 0.73, and using the green channel can increase system performance by 3.51%.