The glmeObject( ) and survObject( ) must be in the following format:
glmeObject <- list(fm, family, par, ran.par, sigma, disp,
lower, upper, str_val, CenObject)
survObject <- list(fm, event, par, disp,
lower, upper, distribution, str_val)
CenObject <- list(fm, family="binomial", par, ran.par,
disp, lower, upper, str_val, Cregime=1, truncated=T, delim_val) | fm | A two-sided linear formula object with the response on the left of a ~ operator and the terms, separated by + operators, on the right. |
| family | A GLM family. |
| par | A character string, naming the parameters. Such as, "alpha", "beta", ... |
| ran.par | A character string, naming the random effects. Such as, "b1","b2",... |
| sigma | A character string, naming the standard deviation for normal distribution. Such as, "sigma". |
| disp | A character vector, specifying the dispersion parameters. |
| lower / upper | lower/upper bounds of the dispersion parameters specified by disp. |
| str_val | A numeric vector of starting values for the fixed parameters in the model. |
| CenObject | A list, indicating the logistic GLME model used to model the censoring mechanism. CenObject=NULL means that the response variable is not censored. |
| event | event indicator |
| distribution | If distribution=NULL, a Cox PH model is fitted. If distribution=weibull, a Weibull model is fitted. |
| delim_val | the lower limit of quantification |
| Cregime | If Cregime=1 (by default), we assume that the censored data are from one regime (point mass). If Cregime=2, we assume that the censored data are from two regimes, one from normal distribution and one from point mass. |
| truncated | logical: if truncated=T (by default), we assume the observed response variable follows a truncated normal distribution; otherwise, we assume it follows a normal distribution. |
See the example in example.md for details.