| Title: | Variable Selection in a Hierarchical Bayesian Model for a Hazard Function |
|---|---|
| Description: | Fits a piecewise exponential hazard to survival data using a Hierarchical Bayesian model with an Intrinsic Conditional Autoregressive formulation for the spatial dependency in the hazard rates for each piece. This function uses Metropolis- Hastings-Green MCMC to allow the number of split points to vary and also uses Stochastic Search Variable Selection to determine what covariates drive the risk of the event. This function outputs trace plots depicting the number of split points in the hazard and the number of variables included in the hazard. The function saves all posterior quantities to the desired path. |
| Authors: | Andrew Chapple [aut, cre] |
| Maintainer: | Andrew Chapple <[email protected]> |
| License: | GPL-2 |
| Version: | 1.1.0 |
| Built: | 2025-03-02 03:17:54 UTC |
| Source: | https://github.com/cran/BayesPieceHazSelect |
PiecewiseBayesSelect
PiecewiseBayesSelect(Y1, I1, X, hyperparameters, beta1start, B, inc, Path, burn)PiecewiseBayesSelect(Y1, I1, X, hyperparameters, beta1start, B, inc, Path, burn)
Y1 |
Vector Containing event times (or censoring time due to death/censoring) |
I1 |
Vector Containing event indicators (1 if l event for a patient, 0 otherwise) |
X |
Matrix of Patient Covariates, the last inc are left out of the selection procedure |
hyperparameters |
List containing 11 hyperparameters and four starting values. In order they are: psi-the swap rate of the SVSS algorithm. c-parameter involved in Sigma matrix for selection. z1a, z1b - beta hyper parameters on probability of inclusion for each of the three hazard functions. a1,b1- hyperparameters on sigma_lambda. clam1- spatial dependency of baseline hazard (between 0 and 1) for the hazard function. Alpha1 - The parameter for the number of split points in the hazard (must be whole number). J1max - Maximum number of split points allowed (must be whole number). J1- Starting number of split points. cl1 -Tuning parameter for log baseline hazard height sampler. |
beta1start |
Starting Values for Beta1 |
B |
Number of iterations |
inc |
Number of variables left out of selection |
Path |
Where to save posterior samples |
burn |
percent of posterior sample to burn in (burn*B must be a whole number) |
##Randomly Generate Semicompeting Risks Data ####Generates random patient time, indicator and covariates. n=100 Y1=runif(n,0,100) I1=rbinom(n,1,.5) library(mvtnorm) X=rmvnorm(n,rep(0,13),diag(13)) ####Read in Hyperparameters ##Swap Rate psi=.5 c=20 ###Eta Beta function probabilities z1a=.4 z1b=1.6 ####Hierarchical lam params ###Sigma^2 lambda_ hyperparameters a1=.7 b1=.7 ##Spacing dependence c in [0,1] clam1=1 #####NumSplit alpha1=3 J1max=10 ####Split Point Starting Value ### J1=3 ##Tuning parameter for lambda cl1=.25 ###Beta Starting Values beta1start=c(0,0,-1,0,0,0,1,1,1,1,1,-1,-1) hyper=c(psi,c,z1a,z1b,a1,b1,clam1,alpha1,J1max,J1,cl1) ###Number of iterations and output location B=200 Path=tempdir() inc=2 burn=.4 PiecewiseBayesSelect(Y1,I1,X,hyper,beta1start,B,inc,Path,burn)##Randomly Generate Semicompeting Risks Data ####Generates random patient time, indicator and covariates. n=100 Y1=runif(n,0,100) I1=rbinom(n,1,.5) library(mvtnorm) X=rmvnorm(n,rep(0,13),diag(13)) ####Read in Hyperparameters ##Swap Rate psi=.5 c=20 ###Eta Beta function probabilities z1a=.4 z1b=1.6 ####Hierarchical lam params ###Sigma^2 lambda_ hyperparameters a1=.7 b1=.7 ##Spacing dependence c in [0,1] clam1=1 #####NumSplit alpha1=3 J1max=10 ####Split Point Starting Value ### J1=3 ##Tuning parameter for lambda cl1=.25 ###Beta Starting Values beta1start=c(0,0,-1,0,0,0,1,1,1,1,1,-1,-1) hyper=c(psi,c,z1a,z1b,a1,b1,clam1,alpha1,J1max,J1,cl1) ###Number of iterations and output location B=200 Path=tempdir() inc=2 burn=.4 PiecewiseBayesSelect(Y1,I1,X,hyper,beta1start,B,inc,Path,burn)