odtr: Function that computes ODTR

In lmmontoya/SL.ODTR: SuperLearner Estimation and Evaluation of ODTR

Function that computes ODTR

Description

Given a W, A, Y dataset, this function will compute the estimated ODTR using SuperLearner. If a QAW function is provided that computes the true E[Y|A,W] (e.g., if simulating), the function will also return the true treatment under the optimal rule and other metrics of evaluating the estimated optimal rule's performance.

Usage

odtr(
  W,
  A,
  Y,
  V,
  rule.output = "d",
  g.SL.library,
  QAW.SL.library,
  blip.SL.library,
  dopt.SL.library = NULL,
  risk.type,
  metalearner,
  kappa = NULL,
  newV = NULL,
  QAW.fun = NULL,
  VFolds = 10,
  grid.size = 100,
  family = NULL,
  ab = NULL,
  cs_to_try = NULL,
  alphas_to_try = NULL
)

Arguments

W	Data frame of observed baseline covariates
A	Vector of treatment
Y	Vector of treatment (continuous or binary)
V	Data frame of observed baseline covariates (subset of W) used to design the ODTR
rule.output	Either "d" for deterministic ODTR or "g" for stochastic ODTR or "rc" for (stochastic) resource-constrained ODTR. Default is "d."
g.SL.library	Character vector for logistic regression modeling the treatment mechanism. Default is 1 (i.e., using mean of A as estimate of g1W).
QAW.SL.library	SuperLearner library for estimating the outcome regression
blip.SL.library	SuperLearner library for estimating blip
dopt.SL.library	SuperLearner library for estimating dopt directly. Default is NULL. Could be "DonV", "Qlearn", "OWL", "EARL", "optclass", "RWL", "treatall", "treatnone". Could also be "all" for all algorithms.
risk.type	Risk type in order to pick optimal combination of coefficients to combine the candidate algorithms. For (1) MSE risk use "CV MSE"; for (2) -E[Ydopt] risk use "CV IPCWDR" (for -E[Ydopt] estimated using double-robust IPTW) or "CV TMLE" (for -E[Ydopt] estimates using TMLE); (3) For the upper bound of the CI of -E[Ydopt] use "CV TMLE CI"
metalearner	Discrete ("discrete"), blip-based ("blip"), vote-based SuperLearner ("vote"). Note that if metalearner is "vote" then cannot put in resource constraints (kappa).
kappa	For ODTR with resource constraints, kappa is the proportion of people in the population who are allowed to receive treatment. Default is NULL.
newV	New V for prediction
QAW.fun	True outcome regression E[Y|A,W]. Useful for simulations. Default is NULL.
VFolds	Number of folds to use in cross-validation. Default is 10.
grid.size	Grid size for simplex.sample() function to create possible combinations of coefficients
family	Either "gaussian" or "binomial". Default is null, if outcome is between 0 and 1 it will change to binomial, otherwise gaussian
ab	Range of Y
cs_to_try	Constants for SL.blip.c
alphas_to_try	Convex combination alphas for SL.blip.alpha

Value

odtr object

References

Luedtke, Alexander R., and Mark J. van der Laan. "Super-learning of an optimal dynamic treatment rule." The international journal of biostatistics 12.1 (2016): 305-332. Coyle, J.R. (2017). Jeremy Coyle, “Computational Considerations for Targeted Learning” PhD diss., University of California, Berkeley 2017 https://escholarship.org/uc/item/9kh0b9vm. Eric Polley, Erin LeDell, Chris Kennedy and Mark van der Laan (2018). SuperLearner: Super Learner Prediction. R package version 2.0-24. https://CRAN.R-project.org/package=SuperLearner.

Examples

## Example
library(SuperLearner)
library(hitandrun)
ObsData = subset(DGP_bin_simple(1000), select = -c(A_star, Y_star))
W = subset(ObsData, select = -c(A,Y))
V = W
A = ObsData$A
Y = ObsData$Y

# blip-based estimate of ODTR with risk function CV-TMLE
odtr(W = W, gform = "W1 + W2", A = A, Y = Y, V = W, blip.SL.library = "SL.blip.HTEepi", QAW.SL.library = "SL.QAW.HTEepi", risk.type = "CV TMLE", metalearner = 'blip')

lmmontoya/SL.ODTR documentation built on Feb. 24, 2024, 10:46 a.m.