摘要
因果推断作为强人工智能的基础,越来越受到科研人员的关注并开始将其应用到各行各业。肺癌是世界范围内癌症发病率和死亡率居高不下的主要原因,患者的生存率低预后差以至于癌症的复发率高。因此确定肿瘤患者预后的因素变得尤为重要。针对传统医学统计方法只能从仅有的观测数据中学习表观现象的建模关联,而没有深入挖掘其隐藏的因果方向,无法回答某些干预和反事实问题,提出基于线性非高斯模型(LiNGAM)的肺癌患者临床数据因果发现方法,可以得出临床病理特征之间的因果路线图。实验结果表明,血小板可以作为肺癌患者预后评估的一个检测指标,从因果推断的角度出发,可以准确判断患者预后,为临床治疗提供有效的干预。为因果推断的应用领域提供了新的研究方向。
As the basis of strong artificial intelligence,causal inference has been paid more and more attention by researchers and has been applied to all walks of life.Lung cancer is the main cause of high cancer incidence and mortality worldwide.The low survival rate and poor prognosis of patients lead to the high recurrence rate of cancer.Therefore,it is quite important to determine the prognostic factors of tumor patients.In view of the fact that traditional medical statistical methods can only learn the modeling association of the apparent phenomena from the only observed data,and can’t answer some intervention and counterfactual,we propose a causal discovery method of lung cancer patients’clinical data based on linear non Gaussian model(lingam),which can obtain the causal roadmap between clinical pathological features.The experiment shows that platelets can be used as a detection index for prognosis evaluation of lung cancer patients.From the perspective of causal inference,it can accurately judge the prognosis of patients,and provide effective intervention for clinical treatment.It provides a new research direction for the application of causal inference.
作者
周琦
万亚平
左建宏
刘纯
马真真
杨菁华
ZHOU Qi;WAN Ya-ping;ZUO Jian-hong;LIU Chun;MA Zhen-zhen;YANG Jing-hua(School of Computers,University of South China,Hengyang 421001,China;Hunan Medical Big Data International Technology Cooperation Base,Hengyang 421001,China;The Third Affiliated Hospital of South China University,Hengyang 421001,China)
出处
《计算机技术与发展》
2021年第8期145-149,共5页
Computer Technology and Development
基金
国防科技创新特区项目(17-163-15-XJ-002-002-04)
湖南省2020年度创新型省份建设专项抗击新冠肺炎疫情应急专题(2020SK3010)
湖南省教育厅重点项目(17A185)
湖南省研究生科研创新项目资助(CX20200936)。
作者简介
周琦(1996-),女,硕士研究生,研究方向为机器学习、因果关系发现;万亚平,博士,教授,硕导,CCF会员(14108M),研究方向为大数据与因果关系;左建宏,博士,教授,硕导,研究方向为肿瘤的诊断、发病机制和防治研究。
