An Associative Memory Approach to Predict Autism Spectrum Disorder

Autism Spectrum Disorder is one of the greatest challenges in diagnosis because it manifests in different forms and does not have a specific medical diagnostic test. With the rapid increase in ASD cases worldwide, the need for accurate and accessible diagnostic tools has increased the manifold. This paper discusses how Bidirectional Associative Memory can be used in the diagnosis of ASD, focusing on its ability to map relationships between input characteristics and diagnostic outcomes. Robustness against noisy and incomplete data creates a unique advantage for BAM in using real-world medical applications. The methodology includes rigorous data preprocessing, feature selection, and bidirectional mapping to help improve the diagnostic accuracy in a way that is interpretable. The model’s applicability and reliability are confirmed using comprehensive evaluation with performance metrics such as accuracy, AUC-ROC, and classification reports. The results thus emphasize that only through integrating innovative models of machine learning with clinical workflows could one possibly facilitate timely intervention and improved patient care for ASD.