SAT.stage2.sampling: Second stage sampling with SAT.

In Penncil/SAT: Surrogate Assisted Two-wave Case Boosting Sampling

Description

This function implements the second stage sampling in SAT by SAT-S or SAT-cY.

Usage

SAT.stage2.sampling(r1, n, S, Rpar = 0.5, r, stage1.index, stage1.y, X, method = "SAT-S")

Arguments

r1	pilot subsample size.
n	total sample size.
S	a binary vector of length n. Surrogate observations for all samples.
Rpar	case proportion parameter. Should be the same as in SAT.stage1.sampling.
r	second stage subsample size.
stage1.index	a vector of length r1. The output of SAT.stage1.sampling, i.e., the index of pilot sampled patients.
stage1.y	a binary vector of length r1. The manual chart review results for patients in stage1.index.
X	a matrix of dimension n times p (the first column needs to be 1). The covariate matrix contains observations for all n samples.

Value

The function returns a list:

beta.pilot	the pilot estimator.
stage1.index	a vector of index for patients who are selected in pilot sampling.
stage2.index	a vector of index for patients who are selected in the second stage sampling.
stage1.weights	a vector of weights used in fitting weighted logistic regression for patients who are selected in pilot sampling.

References

Liu, X., Chubak, J., Hubbard, R. A. & Chen, Y. (2021). SAT: a Surrogate Assisted Two-wave case boosting sampling method, with application to EHR-based association studies.

Examples

library(SAT)
set.seed(0)

colnames(lung_cancer)
X <- cbind(1, lung_cancer[,3:5])
Y <- lung_cancer[,1]
S <- lung_cancer[,2]

# pilot sampling
stage1.index <- SAT.stage1.sampling(r1 = 400, n = 1e5, S, Rpar = 0.5)
# true phenotype collection
stage1.y <- Y[stage1.index]

# second stage sampling
stage2 <- SAT.stage2.sampling(r1 = 400, n = 1e5, S, Rpar = 0.5, r = 800,
                              stage1.index, stage1.y, X, method = "SAT-S")
# true phenotype collection
stage2.y <-  Y[stage2$stage2.index]

stage2$beta.pilot

Penncil/SAT documentation built on Dec. 18, 2021, 7:38 a.m.