Leveraging Light GBM, MLP, and Extra Tree Classifiers for Early Detection and Prediction of Autism Spectrum Disorder

Autism Spectrum Disorder (ASD) is a lifelong neurodevelopmental disorder characterized by challenges in social interaction and communication. Symptoms typically manifest in early childhood and persist into adolescence and adulthood. With the increasing utilization of machine learning techniques in medical research, this paper explores the potential of Light GBM, Multilayer Perceptron (MLP), and Extra Tree Classifiers for predicting and analyzing ASD across different age groups. The study evaluates these methodologies using publicly available datasets representing ASD screening in children, adolescents, and adults. The first dataset, focused on ASD screening in children, comprises 292 instances with 21 attributes. The second dataset, pertaining to ASD screening in adult subjects, includes 704 instances with 21 attributes. The third dataset, related to ASD screening in adolescent subjects, consists of 104 instances with 21 attributes. After employing Light GBM, MLP, and Extra Tree Classifiers and addressing missing values, the results indicate promising performance across all datasets, providing valuable insights into the predictive capabilities of these methodologies for ASD screening in different age groups.