A retrospective study analyzed data from 2152 patients with malignant tumors complicated by sepsis, hospitalized in Guangdong Provincial Hospital of Chinese Medicine from January 2014 to June 2014. The aim was to identify factors affecting hospital survival and develop a predictive model. Multivariate analysis showed that age, SOFA score, coagulation dysfunction and metabolic abnormalities were independent risk factors. The data was divided into a training and test set in a ratio of 8:2, the ADASYN oversampling technique was used, and the selection of key features by the recursive feature elimination (RFE) method was used. Eight machine learning models were evaluated, with random forest performing best with an AUC of 0.95, sensitivity of 91%, and specificity of 85% on the validation set. The RFE method selected the 10 most significant features: troponin T, platelet distribution width, neutrophil count, red blood cell distribution width, fibrinogen, prothrombin time activity, aspartate transaminase, urea, low-density lipoprotein cholesterol, and creatinine. The random forest model provides good predictive power to assess the prognosis of these patients.