International Journal of Scientific Research and Engineering Development

Autism Spectrum Disorder (ASD) in children is a neurodevelopmental condition characterized by difficulties in social interactions, communication, and behaviour. Early detection and diagnosis of ASD, particularly between the ages of 20 and 60 months, are crucial for effective intervention. If not identified early, treatment becomes significantly more challenging. While various machine learning (ML) methods have been applied to predict ASD, the accuracy of predictions for younger age groups remains limited. This paper explores the uses of three machine learning algorithms—Support Vector Machine (SVM), Random Forest, and AdaBoost—to predict and detect autism in children. The AdaBoost classifier, which combines multiple weak learners to create a stronger classifier, is proposed as the primary method. To evaluate the performance of these algorithms, we calculate key metrics such as accuracy, precision, F-score, and the confusion matrix. The algorithm yielding the highest accuracy is then used to predict autism in children.