Prediction of the Breast Cancer Mortality Rate and Its Effective Factors Using Genetic Algorithm and Logistic Regression

Introduction: Logistic regression is one of the most common models used to predict and classify binary and multiple state responses in medicine. Genetic algorithms search techniques inspired by biology have recently been used successfully as a predictive model. The aim of present study was to use the genetic algorithm and logistic regression models in diagnosing and predicting factors affecting breast cancer mortality. Methods: Data of 2836 people with breast cancer during the years 2014-2018 were examined. Information was registered in the cancer registration system of Kerman University of Medical Sciences. Death status was considered as the dependent variable, while age, morphology, tumor differentiation (grad), residence status, and place of residence were considered as independent variables. Sensitivity, specificity, accuracy, and area under the receiver operating characteristic (ROC) curve were used to compare the models. Results: The logistic regression model determined factors affecting the breast cancer mortality rate, (with sensitivity (0.60), specificity (0.80), area under the ROC curve (0.70), and accuracy (0.77)), and also genetic algorithm model (with sensitivity (0.21), specificity (0.96), area under the ROC curve (0.58) and accuracy (0.87)) did so. Conclusion: The sensitivity and area under the ROC curve of the logistic regression model were higher than those of the genetic algorithm, but the specificity and accuracy of the genetic algorithm were higher than those of the logistic regression. According to the purpose of the study, two models can be used simultaneously.