Basis Of The Application Of Cox's Model And The Study On Its Expanding Model

Survival analysis is widely used in the estimation of the treatment effect and the analysis of prognostic factor. The characteristics of survival data are: (1) The distributions of life are varied and difficult to determine. (2) Censored data frequently exist so that they are not so easy to cope with. (3) There are numerous prognostic factors usually hard to control due to the long duration of observation. Among various currently available methods of survival analysis, the non-parametric methods can only solve the first two problems, while the parametric method does for the last two, so that there exists a greater limitation to these two types of methods in application. It is not until 1972 D. R. Cox, a british biostatistician who developed a semiparametric method -- Cox's model that could solve the above three problems simutaneously and idealy, enabled survival analysis to achieve a breakthrough and become a primarily perfect system. Cox's model is the most important mathematical model in the survival analysis nowadays, and it's more and more worldwide application but there still lacks study on some basic problems related to applications.The most domestic articles on survival analysis still adapt the "direct method" to calculate survival rate, which has been done away with gradually since mid-1960's by alien scholars. And there does be some applying the Kplan-Meier Estimation or Log-Rank Test to estimate the survival curve or test the factor effect in terms of univariate analysis. However,the multivariate analysis with Cox's model still rarely happen although it has...