PHD proposal presentation

Date and time: 
Thu, 2016-03-03 11:45
Location / Venue: 

RM 135

PhD Topic: Multivariate copula theory on analysis of survival data with cluster dependencies and repeated measurements

Candidate:  Chrispine Otieno Otieno.


  1. Prof. Patrick Weke (UoN).
  2. Dr. Owuor Nelson.


Multivariate failure time data often arise in medical and epidemiological research and occurs when a sample of survival data consists of clusters and each cluster contains several correlated failure times. Though a wide range of statistical methods are already available to analyze such multivariate survival data, one commonly used statistical method is the copula model which splits the joint survival distribution into two parts: one for the marginal survival functions and the other for the dependence structure of survival times.

Multivariate survival data may arise for instance, from data in disease epidemiology when the hazard rates of infection within a cluster (e.g. a household) are significantly higher for members within one cluster than across clusters. Another common scenario is the case when clinical trials are conducted in multi centers, and patients in one clinical trial center exhibit associated survival probabilities. In general epidemiology, the pertinent question is often to assess the efficacy of treatment methods/vaccination policies across study groups, but this may be confounded by possible correlation between survival outcomes within the clusters.

The copula theory requires that one specifies the joint distribution of the marginal survival probabilities and be able to estimate the dependence and marginal parameter(s). Copulas are therefore useful for constructing families of multivariate survival function with given marginal probability distributions. This study aims at using copulas to model correlated survival data under various scenarios, such as unrestricted dependence parameters, parametric and/or semiparametric marginal distributions among others.  The methods developed shall be validated subject to availability of real time data. 

PhD Abstract 

Expiry Date: 
Sun, 2030-03-03 11:45

UoN Website | UoN Repository | ICTC Website

Copyright © 2017. ICT WebTeam, University of Nairobi