inter_frailty: Fits cure rate frailty model for interval censored data
In JBrettas/intercure: Cure Rate Estimators for Interval Censored Data

Description Usage Arguments Value Examples

View source: R/inter_frailty.R

inter_frailty provides a fit using the frailty model for interval censored data with a cure fraction. Returns a list with the estimated parameters par and their asymptotic covariance matrix mcov.

inter_frailty(dataset, left, right, delta, cov_theta, cov_beta, M, b = 0.001,
  tol = 0.001, max_n = 100, par_cl = NULL, burn_in = 30,
  output_files = FALSE, input_alpha = NULL, input_sigma_alpha = NULL,
  input_cumrisk = NULL)

`dataset`	Dataset used to fit the model.
`left`	Vector containing the last check times before event.
`right`	Vector containing the first check times after event.
`delta`	Flag vector indicating failure inside interval.
`cov_theta`	String vector containing the column names to be used on the cure rate predictor.
`cov_beta`	String vector containing the column names to be used on the predictor associated with the hazard function.
`M`	Number of replicates generated by each iteration on the ANDA (Asymptotic Normal Data Augmentation) algorithm.
`b`	Parameter for initial theta and beta variances.
`tol`	Numeric for tolerance of convergence.
`max_n`	Maximum number of algorithm's iterations without the burn in.
`par_cl`	Registered SOCK cluster for parallel process. If NULL (default) the program loops are executed sequentially.
`burn_in`	Number of burn in iterations.
`output_files`	Boolean indicating if text outputs for the estimates and variances should be generated.
`input_alpha`	Initial parameter vector. Default adopts 0 for each effect. Can be used to increase M for a same fit.
`input_sigma_alpha`	Initial Initial covariance matrix for alpha. Default is an identity matrix multiplied by `b`. Can be used to increase M for a same fit.
`input_cumrisk`	Initial baseline cumulative risk function. Can be used to increase M for a same fit.

The inter_frailty function returns an list containing the following outputs:

`n_its`	number of iterations used on the process.
`par`	estimates of theta and beta parameters.
`mcov`	estimates for the covariance matrix of theta and beta parameters.
`theta_max_dif`	maximum of absolute difference between the effects of the last and the previous iterations.
`stop_c`	stop criteria indicator assuming 1 when process is stopped for a non-convergence criteria. Assumes 0 when convergence is reached.
`last_naalen`	estimated baseline cumulative risk function of the last iteration. Should be stored for fits variating M.

## few iterations just to check how to use the function
set.seed(3)
sample_set <- sim_frailty(80)

inter_frailty(sample_set, sample_set$L, sample_set$R, sample_set$delta,
c("xi1","xi2"), c("xi1","xi2"), M = 10, max_n = 3, burn_in = 0)

## precise estimate (computationally intensive)
## Not run: 

inter_frailty(sample_set, sample_set$L, sample_set$R, sample_set$delta,
c("xi1"), c("xi2"), M = 50, max_n = 100, burn_in = 10)

## End(Not run)