A Comprehensive Review of Risk Prediction in Financial Management of Listed Companies Based on Optimized Deformable Graph Convolutional Networks Under Digital Economy

Abstract

The digital economy has significantly increased the complexity and interconnectedness of financial systems, creating new challenges for risk prediction in publicly listed companies. Traditional statistical models often fail to capture nonlinear dependencies, temporal dynamics, and relational structures inherent in modern financial data, necessitating advanced deep learning approaches. This review explores graph-based deep learning methods, with a focus on Deformable Graph Convolutional Networks (DeformGCNs) for financial risk prediction. Unlike conventional models, graph neural networks capture relationships among entities such as firms, sectors, and markets, enabling improved modeling of interconnected financial systems. DeformGCNs enhance this capability by introducing adaptive receptive fields, allowing dynamic adjustment of node relationships based on evolving market conditions.The study also examines optimization strategies including attention mechanisms, adaptive learning rates, graph sparsification, and multi-scale feature fusion to improve model performance and generalization. These techniques enable effective handling of complex, high-dimensional financial datasets. Empirical evaluations across datasets such as stock exchange records and financial distress benchmarks demonstrate superior predictive accuracy compared to traditional and standard graph-based methods.Despite advancements, challenges such as model interpretability, dynamic graph construction, and real-world deployment remain.This review provides insights into emerging methodologies and future directions for developing robust, scalable, and intelligent financial risk prediction systems in the digital economy.