Clinical Assessment Methods for Early Autism Spectrum Disorder in Paediatrics Using Explainable Artificial Intelligence and Quantum Machine Learning

Abstract

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder characterised by persistent social communication difficulties together with restricted and repetitive behavioural patterns. Early and accurate diagnosis is critical for timely intervention,however, traditional clinical assessment methods require extended time and extensive resources while assessment results depend on evaluator judgement. The development of Artificial Intelligence (AI) technologies including Machine Learning (ML) and Deep Learning (DL) has made it possible to automatically identify ASD through the analysis of EEG data and neuroimaging information and eye-tracking results and behavioural signals. The “black-box” characteristics of these models create two main problems which reduce their effectiveness as predictive tools for medical applications. The introduction of Explainable Artificial Intelligence (XAI) techniques, including SHAP, LIME, and Grad-CAM, provides a framework for improving model interpretability and transparency. Quantum Machine Learning (QML) presents two main benefits through its ability to process high-dimensional data while showing improved performance in computational tasks. This review examines the principal datasets, preprocessing techniques, feature extraction methods, and AI-based detection models used in ASD diagnosis.However, researchers continue to study three main problems which include standardised biomarker deficiencies, data variability and clinical validation shortages. The article presents future research paths which will lead to systems that achieve interpretability and robustness while maintaining clinical usability.