Current Status and Prospects of Combination Drug and Its Artificial Intelligence Algorithm Development

Traditional single-target，single-molecule therapies have shown clear limitations in the treatment of complex diseases，often yielding insufficient efficacy and rapid resistance. Increasing evidence indicates that such diseases are driven by the coordinated dysregulation of multiple signaling pathways，making combination strategies an essential approach. Clinically，rationally designed combinations have demonstrated significant benefits：Dual BRAF/EGFR inhibition enhances objective response rates through synthetic lethality，SHP2/MEK co-inhibition overcomes KRAS mutation—driven resistance，and PD1/CTLA4 blockade markedly prolongs survival in metastatic melanoma. Advances in artificial intelligence （AI） are accelerating combination drug discovery. Leveraging large-scale databases such as DrugComb，predictive models like XGBoost achieve high accuracy （AUROC 0.83 across 22 000 samples），while systems such as DECREASE reduce experimental dose—response testing by over 80% without compromising precision，substantially lowering time and cost. Beyond multi-target applications，innovative strategies are also emerging at the single-target level. For example，orthosteric—allosteric dual-site modulation of PPARγ synergistically enhances efficacy while mitigating side effects，offering new directions for metabolic disease interventions. Immunotherapy has become a leading clinical frontier，with nearly half of approved immune checkpoint inhibitor indications involving combination regimens—either with chemotherapy，targeted therapies，or dual checkpoint blockade—underscoring the value of immune co-regulation. Looking forward，the development of combination therapies requires multidisciplinary integration. Knowledge graph—based approaches can uncover novel synergistic mechanisms，while multi-omics—driven computational platforms combining structural biology，pharmacology，and clinical data will enhance model generalizability. Addressing challenges such as data heterogeneity and limited interpretability through standardized frameworks will be critical for translation from high-throughput prediction to mechanistic insight. The convergence of AI and experimental medicine promises to shorten development cycles，reduce clinical attrition，and drive the evolution of combination drug discovery toward a precision- and mechanism-based paradigm.