Detection of Autism Spectrum Disorder

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder with complex characteristics of social interaction, communication, and behavior difficulties. Early identification of ASD is important for early intervention and successful treatment. The purpose of this project is to create a multimodal system for early detection and classification of ASD using both textual information from behavioral screening questionnaires and visual information from facial images. The system makes use of a machine learning-driven decision tree model for analyzing text responses and a deep learning-driven Convolutional Neural Network (CNN) for processing facial images. The output from both models is fused by employing fuzzy logic in order to come up with a final ASD risk classification. Integration of both text and image information allows the system to present an improved, overall assessment of the likelihood of a person having ASD. Using sophisticated data processing methods, such as feature extraction, normalization, and time frame creation, the system guarantees resilience and flexibility across different applications. This strategy provides an all-inclusive solution to early detection, of ASD, assisting medical specialists in making appropriate choices and enabling prompt intervention to improve outcomes.