Exploring Machine Learning Approaches for Autism Diagnosis through EEG Signal Analysis

Abstract

This paper investigates the potential use of machine learning techniques in the analysis of Electroencephalography (EEG) time series signals to classify Autism Spectrum Disorders (ASD). Our methods ensure that we achieve exceptional accuracy while keeping the method computationally efficient through effective artifact removal algorithms, meaningful feature ex- traction, and powerful machine learning ensemble approaches. EEG data from 28 ASD patients and 28 neurotypical controls were used, and several classification models were evaluated individ- ually and in ensemble combinations. Our proposed method achieved a classification accuracy of 98.85%, demonstrating the proposed approach’s effectiveness in distinguishing ASD using lim- ited EEG datasets. Clinically, this high accuracy displays the method’s potential as a valuable tool for the precise diagnosis of ASD.