Targeted drug design and development, as a core area of modern pharmaceutical research, critically depends on the assessment of protein site druggability as a fundamental component. This review systematically examines the latest research progress and application prospects of drug synergy and antagonism prediction methods that integrate protein three-dimensional spatial structure with artificial intelligence (AI) technologies. This review showcases the molecular biological mechanisms of drug synergism vs antagonism mediated by transcription factors, signal pathway regulation, and membrane transport proteins, and subsequently delves into the molecular structural basis of protein-drug interactions, including precise identification methods for drug binding sites, optimization strategies for molecular docking techniques, and the mechanisms and structural characteristics of multi-target drugs. The review systematically evaluates the practical application progress of AI technologies, especially machine learning and deep learning algorithms, in predicting drug synergy-antagonism effects, as well as the methodological approaches for constructing and evaluating the performance of AI prediction models that integrate multi-source biological data. These research findings provide a solid theoretical foundation for the precision treatment of cancer, infectious diseases, and metabolic disorders, with significant clinical and translational implications for advancing personalized medicine strategies in clinical practice and facilitating the rational design and development of novel multi-target drugs.