Objective: Based on drug therapy of cancer, the paper attempts to evaluate the effect of CHS-DRG grouping, study the methods to optimize the grouping of cases, and provide a basis for the rational formulation of DRG grouping. Methods: A total of 27268 cases related to malignant tumor drug treatment since the payment of CHS-DRG reform from a tertiary cancer specialized hospital in Tianjin were collected. The grouping effect was evaluated with indicators such as coefficient of variation (CV) and reduction in variance (RIV). The internal structural changes of hospitalization expenses were evaluated with methods such as degree of structural change. The influencing factors of hospitalization expenses were screened through the regression analysis, and based on the decision tree E-CHAID method, we construct an optimized grouping plan for malignant tumor drug treatment and determine the payment standard. Results: The efficacy of CHS-DRG grouping in the treatment of malignant tumors was not good enough. By constructing a decision tree model, the factors and classification nodes that significantly affect the hospitalization cost of patients were identified according to the combination therapy, length of hospital stay, application of new technologies, age, and the presence of liver, kidney, and heart function damage. Eight new groups and their hospitalization cost standards were constructed. Conclusion: Based on the decision tree model, cases can be reasonably grouped and objectively reflect the consumption of medical resources, which can provide a reference for adjusting the DRG grouping of malignant tumor drug therapy in the future and promoting the reform of outpatient malignant tumor daytime chemotherapy payment methods.