机器学习预测乳腺癌新辅助治疗后炎症代谢状态改变的模型评价

吴其蓁, 刘启明, 柴烨子, 陶政宇, 王依楠, 郭欣宁, 姜萌, 卜军

Evaluation of machine learning prediction of altered inflammatory metabolic state after neoadjuvant therapy for breast cancer

WU Qizhen, LIU Qiming, CHAI Yezi, TAO Zhengyu, WANG Yinan, GUO Xinning, JIANG Meng, PU Jun

表3 乳腺癌新辅助治疗前后患者的5种模型在测试集中的多特征变量预测性能比较

Tab 3 Comparison of the predictive performance of the five models for multi-feature variables in the test set for patients with breast cancer before and after neoadjuvant therapy

Multi-feature Model AUC Accuracy Precision rate Recall F1 value WBC+HB+HDL RF 0.928 0.914 0.938 0.909 0.923 GB 0.900 0.776 0.885 0.697 0.780 SVM 0.941 0.897 0.909 0.909 0.909 DT 0.799 0.793 0.862 0.758 0.806 KNN 0.907 0.897 0.886 0.939 0.912 IL-2R+IL-8 RF 0.772 0.776 0.763 0.879 0.817 GB 0.792 0.793 0.818 0.818 0.818 SVM 0.764 0.707 0.690 0.879 0.773 DT 0.714 0.707 0.735 0.758 0.746 KNN 0.762 0.724 0.730 0.818 0.771 All features RF 0.954 0.897 0.909 0.909 0.909 GB 0.953 0.914 0.967 0.879 0.921 SVM 0.941 0.879 0.882 0.909 0.896 DT 0.887 0.862 0.857 0.909 0.882 KNN 0.939 0.862 0.903 0.848 0.875