Early Detection of Autism Spectrum Disorder Using Machine Learning

Autism Spectrum Disorder is a neuro-developmental condition that affects multiple aspects of an individual's daily life. Early detection and intervention can significantly reduce its impact. In this paper, we propose a framework to evaluate the performance of key machine learning algorithms for early ASD detection. We apply four feature scaling techniques : Quantile Transformer , Power Transformer , Normalizer, and Max Abs Scaler , to standardize input features. These scaled datasets are then classified using four ML algorithms: AdaBoost , Random Forest , Support Vector Machine, and Linear Discriminant Analysis . The framework is tested on four publicly available ASD datasets representing different age groups: Toddlers, Children, Adolescents, and Adults. The experiments offer a detailed comparison of model performance, identifying the most effective combinations of feature scaling methods and classifiers for accurate ASD detection across age groups, enhancing the potential for improved early diagnosis