StratifiedMedicine (version 1.0.5)

param_est: Parameter Estimation: Across Subgroups

Description

For each identified subgroup, obtain point-estimates and variability metrics (est, SE, CI). fit separate linear regression models. Point-estimates and variability metrics in the overall population are obtained by aggregating subgroup specific results (adaptive weighting or sample size weighting).

Usage

param_est(
  Y,
  A,
  X,
  param,
  mu_hat = NULL,
  Subgrps,
  alpha_ovrl = 0.05,
  alpha_s = 0.05,
  combine = "SS",
  ...
)

Arguments

Y

The outcome variable. Must be numeric or survival (ex; Surv(time,cens) )

A

Treatment variable. (Default supports binary treatment, either numeric or factor). "ple_train" accomodates >2 along with binary treatments.

X

Covariate space.

param

Parameter estimation and inference function. Based on the discovered subgroups, estimate parameter estimates and correspond variability metrics. Options include "lm" (unadjusted linear regression), "dr" (doubly-robust estimator), "gcomp" (G-computation, average the patient-level estimates), "cox" (cox regression), and "rmst" (RMST based estimates as in survRMST package). Default for "gaussian", "binomial" is "dr", while default for "survival" is "cox". Currently only available for binary treatments or A=NULL.

mu_hat

Patient-level estimates (see ple_train)

Subgrps

Identified subgroups. Can be pre-specified, or determined adaptively (see submod_train).

alpha_ovrl

Two-sided alpha level for overall population

alpha_s

Two-sided alpha level at subgroup

combine

Given identified subgroups and correspond point-estimates/SEs/sample sizes, combine="SS" will use sample size weighting for estimates at the overall level. Not applicable for param="dr","ple".

...

Any additional parameters, not currently passed through.

Value

Data-set with parameter estimates and corresponding variability metrics, for overall and subgroups. Subgrps="ovrl" corresponds to the overall population by default.

• param.dat - Parameter estimates and variability metrics (est, SE, LCL/UCL = lower/upper confidence limits, pval = p-value).

References

Funk et al. Doubly Robust Estimation of Causal Effects. Am J Epidemiol 2011. 173(7): 761-767.

Andersen, P. and Gill, R. (1982). Cox<U+2019>s regression model for counting processes, a large sample study. Annals of Statistics 10, 1100-1120.

Uno et al. Moving beyond the hazard ratio in quantifying the between-group difference in survival analysis. Journal of clinical Oncology 2014, 32, 2380-2385.

See Also

param_combine

Examples

# NOT RUN {
library(StratifiedMedicine)

## Continuous ##
dat_ctns = generate_subgrp_data(family="gaussian")
Y = dat_ctns$Y
X = dat_ctns$X
A = dat_ctns$A

## Identify Subgroups: MOB (lmtree) ##
res_lmtree = submod_train(Y, A, X, submod="lmtree")

## Parameter-estimation ##
param.dat = param_est(Y, A, X, param="lm", Subgrps = res_lmtree$Subgrps.train)
param.dat

# }