Reconstruct a trajectory of information accrual
Source:R/information_trajectory.R
information_trajectory.RdWhile estimate_information() provides an estimated information
level at a single point, investigators may want to visualize a trajectory
of how quickly information is accruing as participants are randomized and
their outcomes are obtained. information_trajectory attempts to
reconstruct the study data available at different points in time, and then
compute the information accrued at these times. Regression models can be used
to smooth these trajectories to provide projections of when the information
will reach pre-specified thresholds.
Usage
information_trajectory(
prepared_data,
monitored_design = NULL,
estimation_function,
estimation_arguments,
correction_function = NULL,
orthogonalize,
bootstrap = TRUE,
n_min = 30,
n_increment = 5,
rng_seed,
control = monitored_analysis_control()
)Arguments
- prepared_data
A prepeared dataset: see prepare_monitored_study_data
- monitored_design
An object of class
monitored_designcreated usinginitialize_monitored_design()- estimation_function
A function whose arguments include a data.frame named
data- estimation_arguments
A list of any additional arguments needed by
estimation_function- correction_function
An optional function to adjust the variance estimate using parameters from
estimation_arguments- orthogonalize
A
logicalscalar:: Should estimates, their covariance, and the resulting test statistics be orthogonalized?- bootstrap
A
logicalscalar: should bootstrap be used?- n_min
A
numericscalar indicating the minimum sample size for the information trajectory- n_increment
A
numericscalar indicating the increment in sample size for calculating the trajectory fromn_minto the current sample size inprepared_data.- rng_seed
Numeric scalar containing the L'Ecuyer pseudorandom number generator seed
- control
A list containing the control arguments for computation, typically created with
monitored_analysis_control()