Objective To investigate the predictive value of energy spectral CT radiomics on the prognosis of patients with color-ectal adenocarcinoma.Methods The clinical data and preoperative CT enhancement images of 140 patients with pathology confirmed colorectal adenocarcinoma were analyzed retrospectively.The patients were divided into a good prognosis group(n=60)and a poor prognosis group(n=80)based on the whether they had no tumor recurrence or metastasis and died from tumors within 2 years after operation.The region of interest(ROI)was drawn on the 70 keV enhancement images in CT venous phase,and all radiomics features were further extracted.All patients were randomly divided into a training set and a test set at a ratio of 6∶4,and the least absolute shrinkage and selection operator(LASSO)method was used to screen the radiomics features.The prognosis prediction models of colorectal adenocarcinoma patients were constructed by support vector machine(SVM)and extreme gradient boosting(XGBoost),respectively.The area under the curve(AUC)of receiver operating characteristic(ROC)curve was performed to analyze the predictive effective-ness of various models comprehensively.The AUC of models were verified by DeLong test.Results The differences in the energy spectral parameters including water concentration(WC),iodine concentration(IC),and Effective-Z(Eff-Z)in the venous phase were statistically significant between the two groups(P＜0.05).The AUC of the SVM model constructed from the radiomics features and the combination features were 0.82 and 0.93,with the accuracy of 0.77 and 0.88.Meanwhile,the AUC of the XGBoost model constructed from the radiomics features and the combination features were 0.90 and 0.98,with the accuracy of 0.86 and 0.93.The AUC,accuracy,sensitivity and specificity of the XGBoost model constructed from the combination of energy spectral CT parameters and radiomics features reached the optimal level,and the differences in AUC were statistically significant by DeLong test(P＜0.05).Conclusion Both the SVM and XGBoost models show higher value in preoperative prediction of prognosis of colorectal adenocarcinoma patients,and the combination model of radiomics features and energy spectral parameters(venous phase)has superior predictive effectiveness.