Early Prediction of Autism Spectrum Disorder by Computational Approaches to fMRI Analysis with Early Learning Technique

Abstract

The neuro imaging developmental classification studies are undergone with small amount of samples from the brain activity samples. It promises the inspiring complications in high dimensional data analysis. Autism prediction methodologies are based on behavioral function alone previously which provides good precision but repossession will be unfortunate. We address those problems for early prediction of autism with neural development modern techniques and compared with older. Moreover, visualization of brain activities is quite important in neuro imaging. We believe in better visualization and classification of neuro images in early month captures and appended of Mullen Scales of Early Learning (MSEL). Functional magnetic resonance imaging (fMRI) is one of the controlling tools for measuring non-invasively measure brain activity and it provides with good resolution. For high resolution of brain activity, fMRI gives better than electro encephalon graph (EEG). Visualization of brain activity very clearly is first step to recognize the faults of autism. We have taken into the account for predicting in early Autism Spectrum Disorder (ASD) with help of multiple behavioral activities and development measures using machine learning algorithm. The prediction methods are examined with mostly many prediction methods start to examine the neuro imaging with ultra-high risk factors. The prediction of ASD is moderate accuracy in 14 month development measures from multiple time points. In this proposed work, Mullen early prediction is appended for early prediction and it is examined with computational approach to fMRI analysis with adaptive functioning classifier for machine learning algorithm. This proposed algorithm provides improved version of classification in machine languages with MSEL and high accuracy with conservative methods.