A Data Driven Game Based Framework For Autism Spectrum Disorder Assessment

Authors: Nirupama K, Vivitha R, Mrs. Babisha A, Mrs. Swagatha J P, Dr. Suma Christal Mary S

Abstract: This project is a Data-Driven Game-Based Framework for the Assessment of Autism Spectrum Disorder proposes the construction of a state-of-the-art framework to measure autism spectrum disorder (ASD) in children by combining machine learning and game-based methods. By incorporating interactive cognitive games, checks and MRI interpretation, this work takes advantage of artificial intelligence in order to enhance the diagnostic accuracy of ASD. At the first stage, information is obtained from different public data, such as behavioural research, neuroimaging repositories and validated ASD diagnostic questionnaires. The data has been gathered after preprocessing so that it is clean and structured in a manner that is ready for future analysis. The framework uses supervised learning algorithms, in particular, deep learning architectures such as VGG16 for the analysis of MRI scans and to discover ASD-related neurological features. Whereas game-based tests (and quizzes) are utilized to assess cognitive abilities, social skills and behavioural patterns. The models are also compared based on their quality (e.g. accuracy, precision, recall) to ensure that they have reliability. This study utilizes diverse data sources, such as facial expressions, cognitive responses, and neuroimaging scans, to improve the accuracy of assessments. The developed system demonstrates strong generalization to unseen data, making it suitable for real-world Autism Spectrum Disorder (ASD) diagnosis. These tools enhance accessibility, enabling early detection and timely intervention. Machine learning aids in analyzing behavioral patterns, improving diagnostic accuracy. Clinicians and caregivers can leverage AI-driven insights to support informed decision-making. Overall, this approach holds promise for advancing ASD assessment and intervention strategies.