Exploring Machine Learning Approaches for Diagnosing Autism Spectrum Disorder: Insights from Clinical and Technological Approaches

This paper explores the state-of-the-art research with machine learning applications for detecting Autism Spectrum Disorder (ASD) using eye tracking, sentiment and behavioral analysis, facial feature analysis, and speech analysis. Eye tracking inspects gaze patterns, eye movement duration, and social attention deficits, providing insights into atypical visual processing and aiding early ASD detection. Sentiment and behavioral analysis identifies characteristic social and repetitive behaviors, while facial feature analysis decodes emotional expressions and nonverbal communication challenges, highlighting differences in social interaction. Speech analysis, on the other hand, detects atypical prosody and linguistic patterns, helping to understand verbal communication difficulties in ASD. These multidisciplinary techniques enhance ASD detection by capturing distinctive social, behavioral, and communication markers. By combining these approaches, machine learning offers more accurate diagnostic tools and therapeutic strategies, improving understanding of ASD’s neurodevelopmental differences and potentially supporting earlier interventions.