【摘要】 针对阿尔茨海默病的早期识别问题，提出了一个基于多模态数据的稀疏超图神经网络模型框架。首先，使用稀疏线性回归模型并以样本为节点分别构建脑区和基因超图；其次，使用多核学习方法构建脑区-基因融合超图；最后，基于融合后的超图构建超图神经网络模型，用于多模态数据下阿尔兹海默病及健康对照组的分类。实验结果表明，使用的方法分类准确率达到85.00%，高于传统的图神经网络和图卷积网络，表明该方法在阿尔兹海默病的早期识别中具有优越的分类性能。更多还原

【Abstract】 Aiming at the early identification of Alzheimer’s disease,this paper proposes a sparse hypergraph neural network model framework based on multimodal data.First,brain region and gene hypergraphs are constructed using sparse linear regression model and with samples as nodes,respectively.Second,brain region-gene fusion hypergraphs are constructed using a multinuclear learning method.Finally,hypergraph neural network models are constructed based on the fused hypergraphs for the classification of Alzheimer’s disease and healthy controls under multimodal data.The experimental results show that the classification accuracy of the method we used reaches 85.00%,which is higher than that of the traditional graph neural network and graph convolutional network,indicating that the method has superior classification performance in the early identification of Alzheimer’s disease.更多还原