数字引领时代  智能开创未来

统信大讲堂|带治愈率的长度偏差和区间删失数据的一类新的似然估计方法

彭英伟博士现为加拿大女皇大学(Queen’sUniversity)公共健康科学系终身教授,公共健康科学系生物统计硕士研究生项目主任,数学与统计系兼职教授,女皇大学癌症研究所癌症护理及流行病研究室的资深生物统计分析师,同时也是泛华统计协会加拿大分会现任主席。在主流统计学和医学杂志上发表了120余篇论文,出版CRC/ChapmanHall专著1本,主持或参与加拿大自然科学与工程研究院(NSERC)和美国国立卫生研究院(NIH)等科研基金项目,担任CanadianJournal of Statistics副主编,MathematicalReviews评论员,并担任加拿大卫生研究院(CIHR)、加拿大自然科学与工程研究院(NSERC)等基金评审委员会成员。研究兴趣和领域包括生存分析,癌症研究中的治愈率模型,纵向数据分析,统计计算,因果分析方法等,与在加拿大,美国,中国,西班牙,台湾等地的学者有广泛的合作,已毕业的硕士和博士生多在美国和中国著名大学担任教授或在知名公司和研究所从事统计分析和研究工作。

报告主题:统信大讲堂|带治愈率的长度偏差和区间删失数据的一类新的似然估计方法

报告主题:Anew likelihood estimation method for length-biased andinterval-censored data with a cure fraction

摘要: Length-biaseddata, a special case of left-truncated data, assume that theincidence of the initial event follows a homogeneous Poisson process.I will introduce length-biased and interval-censored data with a curefraction arising from an early-onset diabetes mellitus study and anew method to analyze such data in this talk. The Cox proportionalhazards model for the survival time of the susceptible individualsand the logistic regression model for the probability of beingsusceptible are employed to model the data. We construct the fulllikelihood function and obtain the nonparametric maximum likelihoodestimates of the regression parameters by employing the EM algorithm.The large sample properties of the estimates are established. Theperformance of the method is assessed by simulations. The proposedmodel and method are applied to the data from the early-onsetdiabetes mellitus study. This is collaborative work with Chyong-MeiChen, Hsin-Jen Chen, and Pao-Sheng Shen.