the CMRL: a User-friendly R Package to Estimate Conditional Mean Residual Life Functions

Authors: Kaiyuan Duan and Sujit K. Ghosh Mentor: Sujit Ghosh Affiliation: Department of Statistics, NC State University Address: 2311 Stinson Drive, 5109 SAS Hall Campus Box 8203 NC State University Raleigh, North Carolina 27695 919-515-2528

The conditional mean residual life (MRL) function is the expected remaining lifetime of a subject or a system given survival past a particular time point and the given a set of predictor values. This function is a very useful metric in reliability and actuarial studies when the right tail of the distribution is of interest and can be more relevant than the popularly used survivor or the hazard function. In earlier studies (e.g., see McLain and Ghosh, 2011) it has been shown that estimating conditional MRL without any underlying parametric assumption can be a computationally challenging problem as it involves numerical integrations to obtain smooth estimators. Given the lack of any dedicated computational tools, we develop an R package that not only allows for non-parametric estimation of conditional MRL functions based on censored data but also allows users to fit popular parametric (e.g., Accelerated Failure Time models) and semiparametric (e.g., Proportional Hazard models). Numerical illustrations are based on extensive simulation studies and also real case studies drawn from medical sciences and actuarial studies. Additionally, user-friendly version of the new R package is also demonstrated with numerous practical examples.

Additional Abstract Information

Presenter: Kaiyuan Duan

Institution: NC State University

Type: Poster

Subject: Mathematics

Status: Approved

Time and Location

Session: Poster 8
Date/Time: Tue 5:00pm-6:00pm
Session Number: 5586