Breast cancer risk prediction model based on C5.0 algorithm for postmenopausal women

Abstract

Breast cancer is one of the most common malignant tumors of women in the world, and it most happen in the elderly women, but in recent years the age of onset has become younger. As we know that, postmenopausal women are the groups with less research on breast cancer, and the characteristics of breast cancer are still to be explored. In this paper, based on the characteristic of 1031 postmenopausal women (⩾43 years old) breast cancer data, a breast cancer risk prediction model based on C5.0 algorithm was constructed and the model was optimized. The experimental results show that: a) Compared with machine learning methods such as neural network and support vector machine, C5.0 algorithm has better performance in constructing breast cancer risk prediction model; b) Costmatrix_C5.0 Model with cost matrix is better than adaptiveboosting_c5.0 model with Adaptive Enhancement algorithm; c) The risk of breast cancer is strongly correlated with post-menopausal hormones, age, age of menopause, history of benign breast disease and age of the first childbearing. This research is a practical application of data mining in the medical field and has certain reference value for the clinical diagnosis of breast cancer.