Authors : Mohammad Saeid Zahedi, Zohreh Mortezania and Mahmoud Makkiabadi
Abstract: Breast cancer usually begins from breast tissue and progresses rapidly. The disease is the most common cancer that women suffer from. With late diagnosis of breast cancer, the likelihood of the relapse of the disease is increased. The earlier breast cancer is diagnosed, the greater the likelihood of successful treatment would be. Also, if cancer is diagnosed in the early stages, the likelihood of the relapse of cancerous tumors is decreased. The presence of various symptoms and features of this disease makes it difficult for doctors to diagnose. The neural network provides the possibility of analyzing patient’s clinical data for medical decision making. The purpose of this study is to provide a model for increasing the accuracy of prediction of breast cancer. In this study, patient’s information has been collected from the standard database of Mortaz Super Specialty Hospital of Yazd. The medical records of 574 patients with breast cancer having a total of 32 features have been investigated. Each patient has been followed for at least one year. In order to provide a model of prediction of breast cancer, particle swarm optimization algorithm and MLP neural network are used. The proposed model was compared with the methods of the nearest neighbor, Naïve Bayes and decision tree. The results show that the prediction accuracy of the proposed model is equal to 0.966. Also, for the methods of Naïve Bayes, decision tree and the nearest neighbor, prediction accuracy is 0.91, 0.929 and 0.913, respectively. In predicting breast cancer, the proposed model includes minimum error rate and maximum accuracy and validity compared to other models. Naïve Bayes method has maximum error rate and minimum accuracy.
Mohammad Saeid Zahedi, Zohreh Mortezania and Mahmoud Makkiabadi, 2020. The Application of Particle Swarm Optimization Algorithm to Increase the Accuracy of MLP Neural Network for Prediction of Breast Cancer. Journal of Engineering and Applied Sciences, 15: 3732-3740.