A Machine Learning Model to Predict the ASD Traits and Challenging Behaviors in Children

Abstract

A neurological disorder, along with several behavioral issues, may be to blame for a child's subpar performance in the academic journey (such as anxiety, depression, learning disorders, and irritability). These symptoms can be used to diagnose children with ASD, and supervised machine learning models can help differentiate between ASD traits and other behavioral traits in children, so that a correct diagnosis can be made. Autism usually shows up in a child within the first two years of life, but it can happen at any time. This research aims to create a model for predicting ASD traits and challenging behaviors in children with a better accuracy rate as compared to previous studies. It has analyzed the collected data and proposed a few computational models to predict the ASD traits vs. non-ASD traits among the datasets, as well as identified the most significant behavioral challenges faced by children with ASD traits compared to children without ASD. Several supervised machine learning techniques were tested in this study for their ability to predict ASD traits and behavioral challenges in children and adolescents (2-17), including logistic regression (LR), random forest (RF), naive Bayes (NB), K-nearest neighbor (KNN) and support vector machines (SVM). In the first stage, we used 2650 cases and 20 attributes related to screening for ASD traits in children. In this study, we optimize features and do other types of preprocessing to make sure that our supervised machine learning models give the best results possible. With an accuracy of 99.75% for child autism spectrum disorder (ASD) data and a superb accuracy of 99.27% with a higher precision rate in the complex behavioral features in the children group, Random Forest clearly outperforms the other methods (2–17). Some of the behavioral challenges (inattention, academic performance, anxiety, depression, and learning disorder) were easily predicted by a supervised model with higher prediction scores as compared to previous studies.