基于机器学习方法的乳腺癌预后因素研究

The Study on Prognostic Factors of Breast Cancer Based on Machine Learning Method

收集乳腺癌的预后状况,通过Logistic回归模型、决策树和随机森林算法,分析比较不同变量对乳腺癌预后的敏感性,并比较特征变量的重要性。使用Logistic回归的多变量分析显示,年龄、N-stage、乳腺癌内分泌治疗、化疗手段、多灶性病灶、化疗与预后有关。随机森林分析的结果得到了各个因素指标的重要性评分,有助于正确选择高风险人群,有利于建立分类系统。通过该结果找到危险因素,建立较高准确率的预测模型,预测乳腺癌患者的生存能力,并可作为制定医疗决策的参考。根据患者的特征制定针对患者不同特征的个体化术后随访及辅助治疗策略,争取最大程度上预防乳腺癌的术后复发。

The prognosis of breast cancer is collected. The sensitivity of different variables to the prognosis of breast cancer is analyzed and compared through Logistic regression model, decision tree and random forest algorithm, and the significances of characteristic variables are compared. The multivariate analysis using Logistic regression shows that age, N-stage, endocrine therapy for breast cancer, chemotherapy, multifocal lesions and chemotherapy are related to prognosis. The importance score of each factor index is obtained from the results of random forest analysis, which can help to correctly select high-risk groups and establish the classification system. Based on the results, the risk factors are found, and a prediction model with high accuracy is established to predict the survival ability of patients with breast cancer, which can be used as a reference for making medical decisions. According to the characteristics of patients, individualized postoperative follow-up and adjuvant treatment strategies for different characteristics of patients are developed to maximize the prevention of postoperative recurrence of breast cancer.