Autism Spectrum Disorder Diagnosis with Neural Networks

Abstract

Autism Spectrum Disorder (ASD) affects the whole life of children and leads their families to seek effective treatment and education. According to the Centres for Disease Control and Prevention, the disorder affects one in every 36 children today. Diagnosing this disease at an early age facilitates the treatment process and enables children to be reintegrated into society. The use of Artificial Neural Networks (ANN), one of the artificial intelligence methods used for prediction, has increased in the field of health in recent years and has become an important tool for early disease diagnosis. In this study, single layer perceptron neural networks were designed for the diagnosis of ASD. Data of 14 different parameters taken from children between 12-36 months of age were used, and as a result of the classification, the accuracy value of the neural network was 99.18%, the sensitivity value was 98.91%, the sensitivity value was 1 and the f1 score value was 99.45%. As a result, it is seen that the perceptron classification algorithm has a very high performance in terms of accuracy, precision, sensitivity and f1 score and successfully discriminates the data. © 2024, Ayandegan Institute of Higher Education. All rights reserved.