Disease Gene Prioritization and Drug-Disease Association Prediction Based on Artificial Intelligence-Driven Dynamic Heterogeneous Graph Neural Networks

This paper systematically reviews the research progress in the field of disease gene prioritization and drug-disease association prediction based on dynamic heterogeneous graph neural networks in recent years. Firstly, it elaborates on the core value of the paradigm shift from static association to dynamic process in this field. Subsequently, it combs the technical evolution path of mainstream methods, analyzes the commonly used datasets and evaluation criteria, and summarizes the performance levels achieved by current methods on mainstream datasets. On this basis, this paper deeply explores the existing problems in the current methods, data and evaluation systems, and proposes a series of feasible solutions and optimization paths. Finally, it looks forward to the future development trends of this field, aiming to provide a systematic reference framework for subsequent research. This article aims to systematically summarize the research progress of disease gene prioritization and drug-disease association prediction based on dynamic heterogeneous graph neural networks from multiple dimensions such as methodological evolution, data support, evaluation system and development trend, so as to provide a reference for theoretical research, model optimization and application in the related field.