权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Development and application of semi-parametric survival analysis for time dependent non-linear prognostic factors

时间依赖性非线性预后因素半参数生存分析的开发和应用

基本信息

批准号：
17300091
负责人：
YANAGAWA Takashi
金额：
$ 8.27万
依托单位：
Kurume University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
2005
资助国家：
日本
起止时间：
2005 至 2007
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-17300091/
关键词：
survival analysis time dependent covariate Spline function functional analysis of variance information criterion repeated measurement cohort study stroke 関数分散分析くり返し測定セミパラメトリック法長期間追跡調査

项目摘要

1. Considering the situation that (i) the proportional hazard assumption employed in the Cox regression analysis does not hold, (ii) Covariates depend on time and the relationship between survival time and prognostic factors are non-linear, we developed a flexible method of the analysis of survival data based on the following ideas; (a) employed the first order Spline regression function for modeling the relationship of survival time and prognostic factors, (b) introduced tensor product to take into account the interaction of prognostic factors, and c introduced new algorithm for selecting to meaningful interaction terms and knots. Those theoretical problems that are arisen in the development are attacked by investigators under the leadership of the head investigator. The results are published in those papers listed below.2. We developed a computer program for the new method and applied it to the data from the Hisayama cohort study to explore the relationship between stroke and covari … More ates. It is shown that the risk factors of the stroke that had non-linear relationship with survival time are detected in addition to the known risk factors. Also their risks on the stroke are assessed. Papers demonstrating the usefulness of the method were presented at several international meetings, and a paper submitted for publication, accepted and in press in the Statistics in Medicine.3. The method above is for the survival data. But it is also useful for studying non-linear relationship of response variable and covariates in non-survival data. To demonstrate it the method was applied to the dada from male junior high school students for elucidate linear and non-linear relationships between reduced visual acuity and its contributing factors including living habits and environment. A paper summarizing the results is submitted for publication in a journal.4. At least one international conference was held in a year to publicize the results that we obtained and had critical review of our research by world leading scientists. Less

1. 考虑到(i) Cox回归分析中采用的比例风险假设不成立，（ii）协变量依赖于时间，生存时间与预后因素之间的关系是非线性的，我们基于以下思路开发了一种灵活的生存数据分析方法；(a)采用一阶样条回归函数对生存时间与预后因素之间的关系进行建模，(b)引入张量积来考虑预后因素之间的相互作用，c引入新的算法来选择有意义的相互作用项和结点。在发展过程中出现的理论问题，在首席研究员的带领下，由研究人员来攻克。研究结果发表在下面列出的论文中。我们为新方法开发了一个计算机程序，并将其应用于Hisayama队列研究的数据，以探索卒中与协方差之间的关系。结果表明，除了已知的危险因素外，还发现了与生存时间呈非线性关系的中风危险因素。他们患中风的风险也会被评估。在几次国际会议上提出了证明该方法有用的论文，并在《医学统计》上提交了一篇论文供出版、接受和出版。以上方法为生存数据。但对于研究非生存数据中响应变量与协变量的非线性关系也很有帮助。为了证明这一点，将该方法应用于男性初中生的数据，以阐明视力下降与生活习惯和环境等影响因素之间的线性和非线性关系。一篇总结结果的论文被提交到期刊上发表。每年至少举行一次国际会议，公布我们取得的成果，并得到世界一流科学家对我们研究的批评。少