As a dynamic and reversible chemical modification, RNA modification plays an essential role in many physiological processes of organisms. Among them, N1-methyladenosine(m \(^{1}\) A) plays an important regulatory role in the stability and translation initiation of RNA, and is associated with the occurrence and development of various diseases. However, using traditional biochemical experimental methods to predict these relationships is not only costly but also demands a long experimental duration. Although the computational model can predict the relationship rapidly, the prediction accuracy of a single model needs to be enhanced. So we propose an ensemble model, GDMRMD, which is based on graph convolutional neural network (GCN), DeepWalk, and Metapath2vec to predict the associations between RNA modifications and diseases. We constructed an m \(^{1}\) A-disease heterogeneous network by integrating m \(^{1}\) As similarity network, diseases similarity network, and the adjacency matrix between m \(^{1}\) As and diseases. By applying these three basic models to the constructed heterogeneous network, GDMRMD performed logistic regression on the obtained results to get the final prediction of the relationships between m \(^{1}\) As and diseases. The experiment results show that GDMRMD performs better than the existing single model in the 5-fold cross validation (AUC = 0.89 and AUPR = 0.88).