Comparison of AdaBoost and logistic regression for detecting colorectal cancer patients with synchronous liver metastasis

Abstract

Synchronous liver metastasis is one of the leading causes of the mortality in colorectal cancer patients. In this study, predictive models based on AdaBoost and logistic regression for detecting colorectal cancer patients with synchronous liver metastasis before operation were built and compared. Information gain method, genetic algorithm and AdaBoost were used for feature selection. The predictive performance of each model was evaluated with 10-fold cross-validation and the area under the receiver operating characteristic (ROC) curves. Four predictive variables were identified: CEA, CA50, tumor location (rectum) and maximum diameter. The influence of missing values was also evaluated and compared using serum biomarkers CEA and CA50. Our results indicate that AdaBoost performs better on data set with missing values, while logistic regression has better sensitivity. Both models could be used to develop a predictive model for colorectal cancer patients with synchronous liver metastasis.