Augmented Behavior Analysis for Children with Developmental Disabilities: Building Towards Precision Treatment

Abstract: Children with autism have a developmental problem that gets worse with time. Children with autism have difficulty communicating and interacting with others, and they also exhibit restricted behavior. If their illness is recognized early on and they receive comprehensive care and therapy, children with autism can enjoy happy, full lives. In wealthy countries, it might be difficult to diagnose autistic children until it is too late. Since there are no specific medical tests for autism, a qualified medical professional must make the diagnosis. Given that youngsters need to be closely monitored, medical experts require plenty of time to identify it. In this study, useless to most people images of children were utilized to identify those with autism using artificial intelligence technologies. For the diagnosis of, we have utilized five different algorithms to analyze the prevalence of autism spectrum disorder (ASD) in children: Multilayer Perceptron (MLP), Random Forest (RF), Gradient Boosting Machine (GBM), AdaBoost (AB), and Convolutional Neural Network (CNN). When comparing different algorithms' categorization outcomes, we found that CNN outperformed other traditional Machine Learning (ML) techniques with an accuracy of 92.31%, outperforming all other algorithms. Hence, we suggested a CNN-based prediction model to detect ASD, particularly in youngsters.

Keywords: Convolutional Neural Network (CNN), autism spectrum disorder and Machine Learning