<p>Dengue fever, a mosquito-borne disease caused by dengue virus (DENV), has become a global health problem, and no FDA-approved drug is currently available. Qingwen Baidu Decoction (QBD) is used to treat the critical phase of dengue fever in China, but its mechanism of action remains unclear. In this work, we integrated bioinformatics analysis, machine learning, and network pharmacology to investigate the possible molecular targets and potential active chemical components of QBD. Common targets between differentially expressed genes from DENV infected samples and predicted targets of QBD were identified by bioinformatics analysis and refined by machine learning algorithms including LASSO, random forest and SVM-RFE. Three core genes, CXCL10, EZH2 and EPHB2 were significantly overexpressed in dengue fever patients, indicating their potential diagnostic and therapeutic value. Single cell transcriptome analysis further revealed that QBD primarily targets dendritic cells, monocytes and macrophages. Immune infiltration analysis using ssGSEA showed that these three core genes were significantly associated with CD4⁺ and CD8⁺ T cell subtypes, suggesting their involvement in host immune regulation. Molecular docking and molecular dynamics simulations identified eight chemical components of QBD as potential active ingredients. Based on these computational predictions, we hypothesize that QBD may exert its therapeutic effects through dual mechanisms, including directly binding to DENV proteins to inhibit viral replication, while also regulating the function of CXCL10 and EZH2 to alleviate DENV-induced inflammatory responses and modulate host immunity. These results provide a theoretical reference for future experimental validation and drug development.</p>