计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (3): 146-151.DOI: 10.3778/j.issn.1002-8331.1811-0086

• 模式识别与人工智能 • 上一篇    下一篇

基于贝叶斯方法的乳腺癌预后分析

杜婧,滕婧,马卞,张奥鑫   

  1. 华北电力大学 控制与计算机工程学院,北京 102206
  • 出版日期:2020-02-01 发布日期:2020-01-20

Breast Cancer Patients Prognostic Analysis Using Bayesian Method

DU Jing, TENG Jing, MA Bian, ZHANG Aoxin   

  1. School of Control and Computer Engineering, North China Electric University, Beijing 102206, China
  • Online:2020-02-01 Published:2020-01-20

摘要: 基于贝叶斯方法研究分析乳腺癌患者的临床病理指标对其预后生存率的影响,并对比直接使用患者阳性淋巴结比率(Lymph Node Ratio,LNR)的局部切检值,以及使用LNR的总体估计值之间的效果差异。采用逻辑回归方法估计患者的总体LNR。之后为了反映各临床病理指标对患者预后的动态影响,基于贝叶斯方法构建动态Cox回归模型进行预后分析,仿真结果表明,使用LNR总体估计值的动态Cox回归模型对数据的拟合效果最好,且该模型相对其他模型而言,对总体生存率的预测准确度最高。

关键词: 乳腺癌, 贝叶斯方法, 预后分析

Abstract: This research investigates the influences of the prognostic factors for breast cancer patients based on Bayesian method. In particular, this paper compares the prognostic differences of employing the clinical tested value of Lymph Node Ratio(LNR) and that of the estimated LNR. The LNR is estimated by logistic regressiom, and then the estimated LNR is combined with other prognostic factors to analyze the patients’ overall survival outcome. In order to analyze the temporal dynamic effects of the prognostic factors for patients’ prognosis, a Bayesian dynamic Cox regression model is proposed. The simulation results show that the Bayesian dynamic Cox regression model with the estimated LNR has the best performance in predicting the overall survival rate, compared to the classic Cox model and the ones using the clinical tested LNR.

Key words: breast cancer, Bayesian method, prognosis analysis