Application of Kolmogorov-Arnold Networks Combined with Graph Convolutional Networks in Alzheimer's Disease Diagnosis

This paper proposes a novel framework that combines Kolmogorov-Arnold Networks (KAN) with Graph Convolutional Networks (GCN) for the early diagnosis and pathological analysis of Alzheimer's Disease (AD). By leveraging the nonlinear dynamics of KAN to model complex brain activity patterns and integrating GCN to process brain network data, this approach enables a deeper exploration of structural and functional changes in the brains of AD patients. Experimental results demonstrate that the proposed method significantly improves the accuracy of AD diagnosis, providing a new perspective for early detection. The combination of KAN and GCN presents a promising tool for advancing the understanding of AD and enhancing diagnostic performance in clinical applications.