Prognostic staging of esophageal cancer based on prognosis index and cuckoo search algorithm-support vector machine

Abstract

Esophageal cancer is a heterogeneous malignant tumor. Considering the impact on the postoperative survival with esophageal cancer patients of the blood indicators, constructing a staging system which is superior to the TNM staging system would be helpful to improve the prognosis of patients. In this paper, the blood indicators of esophageal cancer patients are analyzed by Lasso algorithm, Receiver Operating Characteristic curve analysis, and Kaplan-Meier survival analysis. Neutrophil count (NEUT) and prothrombin time (PT) are found to be related to postoperative survival of esophageal cancer patients. Based on TNM stages, NEUT, and PT, the TNM-NPT esophageal cancer prognosis model is established by multiple logistic regression method. The established TNM-NPT prognostic model is superior to the TNM stages in predicting the survival rate of patients with esophageal cancer. The TNM-NPT prognostic staging system is constructed by ROC curve, and TNM-NPT stages are proved to have great classification accuracy by Kaplan-Meier survival analysis. The TNM-NPT prognostic staging system is well predicted by Cuckoo search algorithm-support vector machine with 98.43% accuracy. Therefore, the constructed TNM-NPT prognostic staging system can be successfully used in future clinical studies of esophageal cancer.