Using Machine Learning to Identify Autism Spectrum Disorder Across All Ages

Autism spectrum disorder (ASD) is a neurodevelopmental condition affecting children's societal and cognitive skills, leading to recurring behaviors, limited interests, communiqué challenges, and difficulties in societal interaction. Early detection of ASD can mitigate its sternness and long-term impact. Federated learning (FL) is a cutting-edge technique useful for precise ASD diagnoses in the early stages or preventing prolonged effects. This article applies FL technique uniquely to detect autism. It trains support vector machine (SVM) and local logistic regression (LR) classifiers for ASD classification in both children and adults. Results from classifiers affected by FL are sent to a central server, where a meta-classifier is trained to identify the most accurate approach for detecting ASD in both children and adults. Four distinct datasets related to ASD patients, each comprising over 700 records of affected individuals, were acquired from various repositories for feature extraction. The suggested model accurately predicted ASD in children with a 97% accuracy rate and in adults with an 85% accuracy rate.