Malaysian Journal of Mathematical Sciences, May 2020, Vol. 14, No. 2
Nonparametric Estimation of a Survival Function with Interval Censored Data
Aljawadi, B. A.
Corresponding Email: baderj@hebron.edu
Received date: 20 March 2019
Accepted date: 4 March 2020
Abstract:
Censored data represent a general problem in many fields, especially in medical survival analysis. Where in some cases the censored data replaced with the exact data, which will distort the real distribution of the data set. The disjoint interval-censored data are a special case of the general form of interval censoring which is found in a variety of applications including grouped data and survey responses. However, little attention has been given to their analysis even though they are a recurrent type of data. In contrast to Turnbull's standard non-parametric method for estimation of survival function in case of disjoint interval-censored data, an alternative approach for the estimation of survival function developed in this study. This approach investigated by optimizing the non-parametric maximum likelihood function without any iterative numerical algorithms, where a simple closed-form solution to the non-parametric maximum likelihood function exist. the advantages of the proposed estimation approach are illustrated using real data set with some other examples.
Keywords: Disjoint interval censoring, survival function, covariates and Hessian matrix
