A support vector machine model for predicting non-sentinel lymph node status in patients with sentinel lymph node positive breast cancer.

This study aimed to investigate the accuracy and feasibility of support vector machine (SVM) modeling in predicting non-sentinel lymph node (NSLN) status in patients with SLN-positive breast cancer. Clinicopathological data were collected from 201 cases with sentinel lymph node biopsy breast cancer and included patient age, tumor size, histological type and grade, vascular invasion, estrogen receptor status, progesterone receptor status, CerbB2 status, size and number of positive SLNs, number of negative SLNs, and positive SLN membrane invasion. Feature vector selection was based on a combination of statistical filtration and model-dependent screening. The arbitrary combination with the smallest p value for SVM input was selected, the predicative results of the model were evaluated by a 10-fold cross validation, and a training model was established. Using SLN-positive patients as a double-blind test set, 85 patients were input into the model to analyze its sensitivity and specificity. The combination with the highest cross-validation accuracy was selected for the SVM model and consisted of the following: the number and size of positive SLNs, the number of negative SLNs, and the membrane invasion of positive SLNs. The training accuracy of the model established with the four variables was 92 %, and its cross-validation veracity was 87.6 %. The accuracy of an 85-patient double-blind test of the SVM model was 91.8 %. In conclusion, this SVM model is an accurate and feasible method for the prediction of NSLN status in SLN-positive breast cancer and is conducive to guide clinical treatment.