All Times ET
Keywords: censoring, covariate, survival
In many complex diseases a patient undergoes various disease stages before reaching a terminal state. This fits a multistate model framework where a prognosis may be equivalent to predicting the state occupation at a future time. With the advent of high throughput genomic assays, a clinician may use such high dimensional covariates for making better prediction of state occupation or disease survival. For such prediction purposes, Cox proportional hazards based high dimensional models have been popularly used. However, the assumptions of these proportional hazards based models can be questionable in certain situations. We will discuss an alternative practical solution to this problem by combining a semiparametric approach, called pseudo value regression, with a latent factor or a penalized regression method. We will explore the predictive performances of these combinations in various high dimensional settings through simulated and real data sets and identify the optimal combination for predicting future disease states.