A NOVEL AUTISM SPECTRUM DISORDER DETECTION USING MULTI-LABEL GRAPH CONVOLUTIONAL NETWORK WITH LABEL ATTENTIVE NEIGHBORHOOD CONVOLUTION

Due to the lack of precise medical testing for autism, such as blood tests to detect the illness, diagnosing autism spectrum disorder (ASD) has proven to be challenging. The prevalence of restrictive and/or repetitive behaviors and difficulties and impairments in social communication are hallmarks of autism spectrum disorders. This behavioral condition has been identified. Doctors assess the child's developmental history and behavior to make a diagnosis. Research results. This research used a hybrid Multi Label-Graph Convolutional Network (ML-GCN) with label-attentive neighborhood convolution to categorize the autism spectrum disorder. It offers a clear and effective graph wrapper module in particular for collecting the local attribute data of a specific node to produce a logical representation of node functioning. Additionally, the homeopathic theory recommends developing a taxonomy for attention-related terms. Furthermore, developed an adaptive graph technique that allows the model to learn the kernel for each layer dynamically and uniquely, allowing the model to acquire more valuable and efficient features. On three frequently used reference datasets, including customized and non-specialized networks, comprehensive tests were conducted to validate the neural network-based approach to multi-label classification.