摘要
目的:建立基于机器学习的蒽环类与环磷酰胺化疗方案引起乳腺癌患者恶心呕吐(chemotherapy-induced nausea and vomiting,CINV)的预测模型,为临床个体化止吐药物的选择提供依据。方法:回顾性收集天津医科大学总医院肿瘤科65例使用蒽环类与环磷酰化疗方案的乳腺癌患者临床信息和CINV数据,采用主成分分析(principal component analysis,PCA)降低特征变量维度,利用支持向量机(support vector machine,SVM)、随机森林(random,RF)、K最近邻(K near neighbor,KNN)、朴素贝叶斯(naive bayes,NB)建立CINV预测模型,并对模型进行评估。结果:采用PCA对10个特征变量降维简化,最终确定6个主成分。SVM模型的准确率为0.813,ROC曲线下面积(area under the ROC curve,AUC)为0.817;RF模型的准确率为0.688,AUC为0.650;KNN模型的准确率为0.625,AUC为0.600;NB模型的准确率为0.813,AUC为0.783。结论:SVM模型的性能优于RF、KNN和NB模型,更适用于接受蒽环类与环磷酰胺化疗方案的乳腺癌患者恶心呕吐的预测。
OBJECTIVE To establish a prediction model based on machine learning for anthracycline and cyclophosphamide chemotherapy regimens cause chemotherapy-induced nausea and vomiting(CINV) in breast cancer patients,so as to provide basis for the selection of clinically individualized antiemetic drugs.METHODS The clinical information and CINV data of 65 breast cancer patients who used anthracycline and cyclophosphoryl chemotherapy regimens in Department of Oncology,Tianjin Medical University General Hospital were retrospectively collected.Principal component analysis(PCA) was adopted to reduce variable dimension.Support vector machine(SVM),random forest(RF),K nearest neighbor(KNN),and naive bayes(NB) were used to construct the prediction model.And the model was evaluated.RESULTS PCA reduced the dimensionality of 10 related factors,and finally obtained 6 principal components.The area under ROC curve(AUC) of the SVM model was 0.817 with an accuracy rate of 0.813.The AUC of the RF model was 0.650 with an accuracy rate of 0.688.The AUC of the KNN model was 0.600 with an accuracy rate of 0.625.The AUC of the NB model was 0.783 with an accuracy rate of 0.813.CONCLUSION The performance of the SVM model is better than the RF,KNN and NB models.The RF model can be used to predict the nausea and vomiting of breast cancer patients who use anthracycline and cyclophosphoryl chemotherapy regimens.
作者
杨翀
张靖悦
张毅
袁恒杰
YANG Chong;ZHANG Jingyue;ZHANG Yi;YUAN Hengjie(Department of Pharmacy,Tianjin Huanhu Hospital,Tianjin 300350,China;Department of Pharmacy,Tianjin Medical University General Hospital,Tianjin 300052,China)
出处
《中国医院药学杂志》
CAS
北大核心
2023年第13期1447-1451,共5页
Chinese Journal of Hospital Pharmacy
基金
国家自然科学基金青年项目(编号:81102447)。
关键词
乳腺癌
机器学习
化疗引起的恶心呕吐
蒽环类
环磷酰胺
bread cancer
machine learning
chemotherapy-induced nausea and vomiting
anthracyclines
cyclophosphamide
