Provide Neural Network Prediction Model for Early Detection of Breast Cancer

Document Type : Original Article


1 Head of Computer Engineering Department, Faculty of Engineering, West Tehran Branch, Islamic Azad University, Tehran, Iran

2 Master of Information Technology Student, majoring in Software Design and Production, Faculty of Engineering, West Tehran Branch, Islamic Azad University, Tehran, Iran



Breast cancer is a malignant mass in which breast tissue cells divide without any control due to genetic disorders such as mutations, chromosomal enhancement, deletion, reorganization, displacement, and recurrence. The diagnosis with breast cancer is very time-consuming. If the disease is diagnosed sooner than five years from the first cell deviation, this will increase the patient's chances of survival from 56% to more than 86%, which is very high. Data Mining is a new method for early detection and prognosis of breast cancer. The way presented in the present study will provide a model for predicting and early detection of breast cancer that will help take a step forward with the help of data mining and neural network techniques. This study examines the neural model for diagnosing breast cancer by data analysis on age, weight, age of onset of menopause, age of menopause, duration of OCP use, period of first pregnancy, family history, exercise, and some months of breastfeeding as inputs and disease variables Breast cancer as the output of the feed neural network with the Levenberg-Marquardt post-diffusion learning algorithm and the study of estimating the accuracy of the models by MSE and RMSE methods determined to the principle of multilayer neural networking algorithm has a good result. Also, according to the sensitivity analysis, it was determined that family history is more important, and less important are the variables of age, weight, age of onset of menstruation, and the number of months of breastfeeding. According to the importance of personal information in early detection of breast cancer, the efficiency of the Internet of Things in smart cities can be exploited