A machine learning approach to predict autism spectrum disorder (ASD) for both children and adults using feature optimization

A central nervous system known as an Autism Spectrum Disorder (ASD) has long-term effects on a person's capacity for engagement and interaction with others. Since its symptoms often manifest in the first two years of life, autism is considered to be a behavioral condition that can be identified at any point in a person's life. This study investigated the potentiality of machine learning techniques such as Logistic Regression, Random Forest, Multinomial Naive Bayes (MNB), Bernoulli Naive Bayes (BNB), Support Vector Machine (SVM), and Gaussian Naive Bayes (GNB) to predict ASD using some health parameters. There are 292 instances and 21 attributes in the first dataset linked to the screening for ASD in children. The adult individuals in the second dataset had a total of 704 occurrences and 21 characteristics related to ASD detection. In order to achieve the highest accuracy possible from the machine learning models, feature optimization is used in this study along with other preprocessing approaches. The findings overwhelmingly support the notion that Random Forest performs better on all of these datasets, with the greatest accuracy (100%) for data on Autistic Spectrum Disorder (ASD) in children and adults, respectively.