Using Different Machine Learning Techniques to Predict Recurrent Breast Cancer at Young Age

A large body of literature shows that postmenopausal women over the age of 50 are at great risk for recurrent breast cancer; however, studies on recurrent breast cancer in younger women have been scarce. This study is to use different machine learning technologies to identify the risk factors and clinical features for breast cancer recurrence. Clinical datasets were a total of 5,788 valid records including 749 recurrent cases. In addition, this study uses the oversampling technique to adjust the imbalance problem. The results showed that the important risk factors for the samples in the two age groups are the same, namely, the pathological, surgical, and clinical stages. The classification and regression trees method showed the highest accuracy in prediction: 0.7907 for those aged < 50 years and 0.8349 for those aged ≥ 50 years. The risk prediction model developed in this study may provide evidence for the robustness of the breast cancer clinical risk prediction model.