params_med_np {vaccine} | R Documentation |
Set parameters controlling nonparametric estimation of mediation effects
Description
This should be used in conjunction with est_med
to
set parameters controlling nonparametric estimation of mediation effects;
see examples.
Usage
params_med_np(
grid_size = list(y = 101, s = 101, x = 5),
surv_type = "survML-G",
density_type = "binning",
density_bins = 15
)
Arguments
grid_size |
A list with keys |
surv_type |
One of c("Cox", "survSL", "survML-G", "survML-L"); controls the method to use to estimate the conditional survival and conditional censoring functions. If type="Cox", a survival function based on a Cox proportional hazard model will be used. If type="survSL", the Super Learner method of Westling 2023 is used. If type="survML-G", the global survival stacking method of Wolock 2022 is used. If type="survML-L", the local survival stacking method of Polley 2011 is used. |
density_type |
One of c("binning", "parametric"); controls the method to use to estimate the density ratio f(S|X)/f(S). |
density_bins |
An integer; if density_type="binning", the number of bins to use. If density_bins=0, the number of bins will be selected via cross-validation. |
Value
A list of options.
Examples
data(hvtn505)
dat <- load_data(time="HIVwk28preunblfu", event="HIVwk28preunbl", vacc="trt",
marker="IgG_V2", covariates=c("age","BMI","bhvrisk"),
weights="wt", ph2="casecontrol", data=hvtn505)
ests_med <- est_med(
dat = dat,
type = "NP",
t_0 = 578,
params_np = params_med_np(surv_type="survML-L")
)