摘要
目的 通过构建和优化基于46个基因的三阴性乳腺癌(TNBC)患者新辅助化疗(NAC)疗效的预测模型,提供一种对患者进行精准分型的方法,更好地促进NAC在TNBC治疗中的应用。方法 从基因表达综合(GEO)数据库的公共数据集(GSE163882)获取221个接受NAC的TNBC患者的全基因组数据,进行模型构建特征基因的筛选和过滤,构建预测TNBC患者NAC疗效的模型。评估不同基因数构建的7种不同机器学习算法对TNBC患者NAC疗效的预测价值,最后利用深度神经网络模型进行优化。结果 对221个患者样本的基因进行模型构建特征基因的筛选和过滤后,获得46个差异表达基因作为最后模型构建特征数。通过7种机器学习模型的测试及优化后发现,线性判别分析模型预测TNBC患者NAC疗效的准确率达到85.00%,后续利用深度神经网络模型优化,可使预测的准确率提高到90.00%,相比于7种模型均有显著提高。结论 基于基因表达数目构建的模型具有预测TNBC患者NAC疗效的作用,可为后期的病情预测提供参考依据。
Objective To provide a method for accurate typing of patients by constructing a prediction model for the efficacy of neoadjuvant chemotherapy(NAC)in triple negative breast cancer(TNBC)based on expression of 46 genes,and to promote the application of NAC in the treatment of TNBC.Method The whole-genome transcriptome informa-tion of 221 breast cancer patients who received NAC was obtained from the public data sets(GSE163882)of Gene Ex-pression Omnibus(GEO).Based on the obtained patient information and data,the characteristic genes of model construc-tion were screened and filtered,and the model predicting efficacy of NAC in TNBC patients was constructed.Evaluate the predictive value of 7 different machine learning algorithms constructed with different gene numbers for the efficacy of NAC in TNBC patients,and finally optimize them using a deep neural network model.Result After screening and filter-ing the characteristic genes of 221 patient samples for model construction,46 differentially expressed genes were ob-tained as the final model construction characteristic values.After testing and optimization the 7 machine learning models,it was found that the accuracy of the linear discriminant analysis model reached 85.00%,and the subsequent optimization of the deep neural network model improved the accuracy of the prediction to 90.00%,which was significantly improved compared with the seven models.Conclusion The model constructed based on gene expression level can predict the effi-cacy of NAC in TNBC patients,which can provide reference for subsequent disease prediction.
作者
谢文倩
庄颖
刘智威
张恋恋
袁沛怡
龚浩
XIE Wenqian;ZHUANG Ying;LIU Zhiwei;ZHANG Lianlian;YUAN Peiyi;GONG Hao(School of Life Science,Huizhou University,Huizhou 516007,Guangdong,China)
出处
《癌症进展》
2023年第15期1646-1650,共5页
Oncology Progress
基金
广东省科技创新战略专项资金项目(pdjh2021b0477)。
关键词
三阴性乳腺癌
传统机器算法
深度神经网络
triple negative breast cancer
traditional machine algorithm
deep neural network