MLEAFT {AmoudSurv}

Accelerated Failure Time (AFT) Model.

Description

Tractable Parametric accelerated failure time (AFT) model's maximum likelihood estimation, log-likelihood, and information criterion. Baseline hazards: NGLL,GLL,MLL,PGW, GG, EW, MKW, LL, TLL, SLL,CLL,SCLL,ATLL, and ASLL

Usage

MLEAFT(
  init,
  times,
  status,
  n,
  basehaz,
  z,
  method = "BFGS",
  hessian = TRUE,
  conf.int = 0.95,
  maxit = 1000,
  log = FALSE
)

Arguments

init

: initial points for optimisation

times

: survival times

status

: vital status (1 - dead, 0 - alive)

n

: The number of the data set

basehaz

: baseline hazard structure including baseline (New generalized log-logistic accelerated failure time "NGLLAFT" model, generalized log-logisitic accelerated failure time "GLLAFT" model, modified log-logistic accelerated failure time "MLLAFT" model, exponentiated Weibull accelerated failure time "EWAFT" model, power generalized weibull accelerated failure time "PGWAFT" model, generalized gamma accelerated failure time "GGAFT" model, modified kumaraswamy Weibull proportional odds "MKWAFT" model, log-logistic accelerated failure time "LLAFT" model, tangent-log-logistic accelerated failure time "TLLAFT" model, sine-log-logistic accelerated failure time "SLLAFT" model, cosine log-logistic accelerated failure time "CLLAFT" model, secant-log-logistic accelerated failure time "SCLLAFT" model, arcsine-log-logistic accelerated failure time "ASLLAFT" model, arctangent-log-logistic accelerated failure time "ATLLAFT" model, Weibull accelerated failure time "WAFT" model, gamma accelerated failure time "GAFT", and log-normal accelerated failure time "LNAFT")

z

: design matrix for covariates (p x n), p >= 1

method

:"optim" or a method from "nlminb".The methods supported are: BFGS (default), "L-BFGS", "Nelder-Mead", "SANN", "CG", and "Brent".

hessian

:A function to return (as a matrix) the hessian for those methods that can use this information.

conf.int

: confidence level

maxit

:The maximum number of iterations. Defaults to 1000

log

:log scale (TRUE or FALSE)

Value

a list containing the output of the optimisation (OPT) and the log-likelihood function (loglik)

Author(s)

Abdisalam Hassan Muse, Samuel Mwalili, Oscar Ngesa, Christophe Chesneau [e-mail]

Examples

#Example #1
data(alloauto)
time<-alloauto$time
delta<-alloauto$delta
z<-alloauto$type
MLEAFT(init = c(1.0,0.20,0.05),times = time,status = delta,n=nrow(z),
basehaz = "WAFT",z = z,method = "BFGS",hessian=TRUE, conf.int=0.95,maxit = 1000,
log=FALSE)

#Example #2
data(bmt)
time<-bmt$Time
delta<-bmt$Status
z<-bmt$TRT
MLEAFT(init = c(1.0,1.0,0.5),times = time,status = delta,n=nrow(z),
basehaz = "LNAFT",z = z,method = "BFGS",hessian=TRUE, conf.int=0.95,maxit = 1000,log=FALSE)

#Example #3
data("gastric")
time<-gastric$time
delta<-gastric$status
z<-gastric$trt
MLEAFT(init = c(1.0,0.50,0.5),times = time,status = delta,n=nrow(z),
basehaz = "LLAFT",z = z,method = "BFGS",hessian=TRUE, conf.int=0.95,maxit = 1000,
log=FALSE)

#Example #4
data("larynx")
time<-larynx$time
delta<-larynx$delta
larynx$age<-as.numeric(scale(larynx$age))
larynx$diagyr<-as.numeric(scale(larynx$diagyr))
larynx$stage<-as.factor(larynx$stage)
z<-model.matrix(~ stage+age+diagyr, data = larynx)
MLEAFT(init = c(1.0,0.5,0.5,0.5,0.5,0.5,0.5,0.5),times = time,status = delta,n=nrow(z),
basehaz = "LNAFT",z = z,method = "BFGS",hessian=TRUE, conf.int=0.95,maxit = 1000,
log=FALSE)