The paper presents a novel model, APP-DGNN, for predicting academic performance in online learning environments. The model leverages dual graph neural networks to effectively utilize both interaction activities and student attributes. The interaction-based graph neural network (IGCN) module learns local academic performance representations from online interaction activities, while the attribute-based graph neural network (AGCN) module learns global academic performance representations from student attributes using dynamic graph convolution operations. The learned representations from both levels are combined in a local-to-global representation learning module to generate predicted academic performances. Empirical results on a widely recognized public dataset demonstrate that the proposed model significantly outperforms existing methods, achieving an accuracy of 83.96% for predicting students who pass or fail and 90.18% for predicting students who pass or withdraw. Ablation studies confirm the effectiveness and superiority of the proposed techniques. The paper also discusses the contributions, related works, and experimental setup, providing a comprehensive overview of the proposed model and its performance.The paper presents a novel model, APP-DGNN, for predicting academic performance in online learning environments. The model leverages dual graph neural networks to effectively utilize both interaction activities and student attributes. The interaction-based graph neural network (IGCN) module learns local academic performance representations from online interaction activities, while the attribute-based graph neural network (AGCN) module learns global academic performance representations from student attributes using dynamic graph convolution operations. The learned representations from both levels are combined in a local-to-global representation learning module to generate predicted academic performances. Empirical results on a widely recognized public dataset demonstrate that the proposed model significantly outperforms existing methods, achieving an accuracy of 83.96% for predicting students who pass or fail and 90.18% for predicting students who pass or withdraw. Ablation studies confirm the effectiveness and superiority of the proposed techniques. The paper also discusses the contributions, related works, and experimental setup, providing a comprehensive overview of the proposed model and its performance.