Recent Advances in Risk prediction in financial management of listed companies based on optimized Deformable graph convolutional networks under digital economy: A Systematic Review

Abstract

The rapid evolution of the digital economy has significantly increased the complexity, volatility, and interconnectedness of financial systems, posing major challenges for risk prediction in publicly listed companies. Traditional statistical and machine learning approaches often fail to capture nonlinear dependencies, temporal dynamics, and relational structures inherent in modern financial data.  This systematic review examines the application of Deformable Graph Convolutional Networks (DGCNs) for financial risk prediction. Unlike conventional graph neural networks, DGCNs introduce adaptive receptive fields that dynamically adjust neighborhood relationships, enabling more accurate modeling of evolving financial networks.The review highlights key optimization strategies, including attention mechanisms, temporal modeling using LSTM and GRU, graph regularization techniques, and integration of knowledge graphs to enhance feature representation. These approaches improve the model’s ability to capture complex dependencies across entities such as firms, markets, and supply chains. Empirical studies across diverse datasets, including stock exchange records, financial statements, and alternative data sources, demonstrate that DGCN-based models outperform traditional and baseline methods in tasks such as credit risk assessment, financial distress prediction, and fraud detection. Despite these advancements, challenges remain in scalability, interpretability, and real-world deployment. This review provides a comprehensive overview of current methodologies and outlines future directions for developing intelligent, robust, and scalable financial risk prediction systems in the digital economy.