The aim of this article was to use the Support Vector Machine (SVM) to predict the benign and malignant solitary pulmonary nodules (SPNs) in early-stage lung cancer in order to lessen the patient’s pain and save the money. Fifty and one patient records were collected .Each record consisted of four clinical characteristics and nine morphological characteristics. The SVM classifier was built by radial basis kernel function. The penalty factor C and kernel parameter σ were optimized by comparing particle swarm optimization (PSO), grid search algorithm (GSA) and genetic algorithm (GA)and then employed to diagnose the SPNs. By comparison with a Logistic regression (LR) model, the overall results of our calculation demonstrated that the area under the receiver operator characteristic (ROC) curve for the model (0.913 ± 0.051, p<0.05) was higher than the LR model. The accuracy, sensitivity and specificity in the model were 90.7%, 89.3% and 93.3% respectively. It is represented that the PSO-SVM model can be used in predicting the early-stage lung nodules.
Li, Shan; Yu, Ying; and Chen, Haibin, "A NEW METHOD FOR PREDICTING EARLY-STAGE LUNG NODULES BASED ON PSO-SVM HYBRID ALGORITHM" (2016). PACIS 2016 Proceedings. 397.