getIPWeights: Inverse Probability Weights.

In osofr/stremr: Streamlined Estimation for Static, Dynamic and Stochastic Treatment Regimes in Longitudinal Data

Description

Evaluate the inverse probability weights for up to 3 intervention nodes: CENS, TRT and MONITOR. This is based on the inverse of the propensity score fits for the observed likelihood (g0.C, g0.A, g0.N), multiplied by the indicator of not being censored and the probability of each intervention in intervened_TRT and intervened_MONITOR. Requires column name(s) that specify the counterfactual node values or the counterfactual probabilities of each node being 1 (for stochastic interventions). The output is person-specific data with evaluated weights, wts.DT, only observation-times with non-zero weight are kept Can be one regimen per single run of this block, which are then combined into a list of output datasets with lapply. Alternative is to allow input with several rules/regimens, which are automatically combined into a list of output datasets.

Usage

getIPWeights(
  OData,
  intervened_TRT = NULL,
  intervened_MONITOR = NULL,
  useonly_t_TRT = NULL,
  useonly_t_MONITOR = NULL,
  rule_name = paste0(c(intervened_TRT, intervened_MONITOR), collapse = ""),
  tmax = NULL,
  ignore_tmin = NULL,
  ignore_tmax = NULL,
  reverse_wt_prod = FALSE,
  holdout = FALSE,
  eval_stabP = TRUE,
  trunc_weights = Inf,
  intervened_type_TRT = NULL,
  intervened_type_MONITOR = NULL
)

Arguments

OData	Input data object created by importData function.
intervened_TRT	Column name in the input data with the probabilities (or indicators) of counterfactual treatment nodes being equal to 1 at each time point. Leave the argument unspecified (NULL) when not intervening on treatment node(s).
intervened_MONITOR	Column name in the input data with probabilities (or indicators) of counterfactual monitoring nodes being equal to 1 at each time point. Leave the argument unspecified (NULL) when not intervening on the monitoring node(s).
useonly_t_TRT	Use for intervening only on some subset of observation and time-specific treatment nodes. Should be a character string with a logical expression that defines the subset of intervention observations. For example, using "TRT == 0" will intervene only at observations with the value of TRT being equal to zero. The expression can contain any variable name that was defined in the input dataset. Leave as NULL when intervening on all observations/time-points.
useonly_t_MONITOR	Same as useonly_t_TRT, but for monitoring nodes.
rule_name	Optional name for the treatment/monitoring regimen.
tmax	Maximum value of the follow-up period. All person-time observations above this value will be excluded from the output weights dataset.
ignore_tmin	(ADVANCED FEATURE) Minimum value of the follow-up period at which the IP-weights should start accumulating over time. All IP-weights for time-points t < ignore_tmin will be set to a constant 1. This will have the effect of completely ignoring all weight contributions that occur before ignore_tmin.
ignore_tmax	(ADVANCED FEATURE) Maximum value of the follow-up period to start accumulative the weights over time. All of the time-specific IP-weights with t < ignore_tmin will be set to constant 1 PRIOR to the evaluation of the cumulative weights. This will have the effect of completely ignoring all the IP weight contributions up to and including the time-point ignore_tmin.
reverse_wt_prod	Set to TRUE to take the product of the cumulative weights in reverse time-ordering. That is, the cumulative product will be evaluated by starting from the highest follow-up time point (time variable value).
holdout	Obtain the weights based on out-of-sample (holdout / validation set) predictions of propensity scores. This is useful for running CV-TMLE or evaluating the quality of the model fits based on validation sets.
eval_stabP	Evaluate the additional weight stabilization factor for each time-point. This is used for MSMs only and is enabled by default.
trunc_weights	Specify the numeric weight truncation value. All final weights exceeding the value in trunc_weights will be truncated.
intervened_type_TRT	(ADVANCED FUNCTIONALITY) Set to NULL by default, can be characters that are set to either "bin", "shift" or "MSM". Provides support for different types of interventions on TRT (treatment) node (counterfactual treatment node A^*(t)). The default behavior is the same as "bin", which assumes that A^*(t) is binary and is set equal to either 0, 1 or p(t), where 0<=p(t)<=1. Here, p(t) denotes the probability that counterfactual A^*(t) is equal to 1, i.e., P(A^*(t)=1)=p(t) and it can change in time and subject to subject. For "shift", it is assumed that the intervention node A^*(t) is a shift in the value of the continuous treatment A, i.e., A^*(t)=A(t)+delta(t). Finally, for "MSM" it is assumed that we simply want the final intervention density g^*(t) to be set to a constant 1. This has use for static MSMs.
intervened_type_MONITOR	(ADVANCED FUNCTIONALITY) Same as intervened_type_TRT, but for monitoring intervention node (counterfactual monitoring node N^*(t)).

Value

A data.table with cumulative weights for each subject and each time-point saved under column "cum.IPAW".

Examples

options(stremr.verbose = TRUE)
require("data.table")

# ----------------------------------------------------------------------
# Simulated Data
# ----------------------------------------------------------------------
data(OdataNoCENS)
OdataDT <- as.data.table(OdataNoCENS, key=c("ID", "t"))

# define lagged N, first value is always 1 (always monitored at the first time point):
OdataDT[, ("N.tminus1") := shift(get("N"), n = 1L, type = "lag", fill = 1L), by = ID]
OdataDT[, ("TI.tminus1") := shift(get("TI"), n = 1L, type = "lag", fill = 1L), by = ID]

# ----------------------------------------------------------------------
# Define intervention (always treated):
# ----------------------------------------------------------------------
OdataDT[, ("TI.set1") := 1L]
OdataDT[, ("TI.set0") := 0L]

# ----------------------------------------------------------------------
# Import Data
# ----------------------------------------------------------------------
OData <- importData(OdataDT, ID = "ID", t = "t", covars = c("highA1c", "lastNat1", "N.tminus1"),
                    CENS = "C", TRT = "TI", MONITOR = "N", OUTCOME = "Y.tplus1")

# ----------------------------------------------------------------------
# Look at the input data object
# ----------------------------------------------------------------------
print(OData)

# ----------------------------------------------------------------------
# Access the input data
# ----------------------------------------------------------------------
get_data(OData)

# ----------------------------------------------------------------------
# Model the Propensity Scores
# ----------------------------------------------------------------------
gform_CENS <- "C ~ highA1c + lastNat1"
gform_TRT = "TI ~ CVD + highA1c + N.tminus1"
gform_MONITOR <- "N ~ 1"
stratify_CENS <- list(C=c("t < 16", "t == 16"))

# ----------------------------------------------------------------------
# Fit Propensity Scores
# ----------------------------------------------------------------------
OData <- fitPropensity(OData, gform_CENS = gform_CENS,
                        gform_TRT = gform_TRT,
                        gform_MONITOR = gform_MONITOR,
                        stratify_CENS = stratify_CENS)

# ----------------------------------------------------------------------
# IPW Ajusted KM or Saturated MSM
# ----------------------------------------------------------------------
require("magrittr")
AKME.St.1 <- getIPWeights(OData, intervened_TRT = "TI.set1") %>%
             survNPMSM(OData) %$%
             estimates
AKME.St.1

# ----------------------------------------------------------------------
# Bounded IPW
# ----------------------------------------------------------------------
IPW.St.1 <- getIPWeights(OData, intervened_TRT = "TI.set1") %>%
            directIPW(OData)
IPW.St.1[]

# ----------------------------------------------------------------------
# IPW-MSM for hazard
# ----------------------------------------------------------------------
wts.DT.1 <- getIPWeights(OData = OData, intervened_TRT = "TI.set1", rule_name = "TI1")
wts.DT.0 <- getIPWeights(OData = OData, intervened_TRT = "TI.set0", rule_name = "TI0")
survMSM_res <- survMSM(list(wts.DT.1, wts.DT.0), OData, tbreaks = c(1:8,12,16)-1,)
survMSM_res$St

# ----------------------------------------------------------------------
# Sequential G-COMP
# ----------------------------------------------------------------------
t.surv <- c(0:10)
Qforms <- rep.int("Qkplus1 ~ CVD + highA1c + N + lastNat1 + TI + TI.tminus1", (max(t.surv)+1))
params <- gridisl::defModel(estimator = "speedglm__glm")

## Not run: 
gcomp_est <- fit_GCOMP(OData, tvals = t.surv, intervened_TRT = "TI.set1",
                          Qforms = Qforms, models = params, stratifyQ_by_rule = FALSE)
gcomp_est[]

## End(Not run)
# ----------------------------------------------------------------------
# TMLE
# ----------------------------------------------------------------------
## Not run: 
tmle_est <- fit_TMLE(OData, tvals = t.surv, intervened_TRT = "TI.set1",
                    Qforms = Qforms, models = params, stratifyQ_by_rule = TRUE)
tmle_est[]

## End(Not run)

# ----------------------------------------------------------------------
# Running IPW-Adjusted KM with optional user-specified weights:
# ----------------------------------------------------------------------
addedWts_DT <- OdataDT[, c("ID", "t"), with = FALSE]
addedWts_DT[, new.wts := sample.int(10, nrow(OdataDT), replace = TRUE)/10]
survNP_res_addedWts <- survNPMSM(wts.DT.1, OData, weights = addedWts_DT)

# ----------------------------------------------------------------------
# Multivariate Propensity Score Regressions
# ----------------------------------------------------------------------
gform_CENS <- "C + TI + N ~ highA1c + lastNat1"
OData <- fitPropensity(OData, gform_CENS = gform_CENS, gform_TRT = gform_TRT,
                        gform_MONITOR = gform_MONITOR)

# ----------------------------------------------------------------------
# Fitting treatment model with Gradient Boosting machines:
# ----------------------------------------------------------------------
## Not run: 
require("h2o")
h2o::h2o.init(nthreads = -1)
gform_CENS <- "C ~ highA1c + lastNat1"
models_TRT <- sl3::Lrnr_h2o_grid$new(algorithm = "gbm")
OData <- fitPropensity(OData, gform_CENS = gform_CENS,
                        gform_TRT = gform_TRT,
                        models_TRT = models_TRT,
                        gform_MONITOR = gform_MONITOR,
                        stratify_CENS = stratify_CENS)

# Use `H2O-3` distributed implementation of GLM for treatment model estimator:
models_TRT <- sl3::Lrnr_h2o_glm$new(family = "binomial")
OData <- fitPropensity(OData, gform_CENS = gform_CENS,
                        gform_TRT = gform_TRT,
                        models_TRT = models_TRT,
                        gform_MONITOR = gform_MONITOR,
                        stratify_CENS = stratify_CENS)

# Use Deep Neural Nets:
models_TRT <- sl3::Lrnr_h2o_grid$new(algorithm = "deeplearning")
OData <- fitPropensity(OData, gform_CENS = gform_CENS,
                        gform_TRT = gform_TRT,
                        models_TRT = models_TRT,
                        gform_MONITOR = gform_MONITOR,
                        stratify_CENS = stratify_CENS)

## End(Not run)

# ----------------------------------------------------------------------
# Fitting different models with different algorithms
# Fine tuning modeling with optional tuning parameters.
# ----------------------------------------------------------------------
## Not run: 
params_TRT <- sl3::Lrnr_h2o_grid$new(algorithm = "gbm",
                              ntrees = 50,
                              learn_rate = 0.05,
                              sample_rate = 0.8,
                              col_sample_rate = 0.8,
                              balance_classes = TRUE)
params_CENS <- sl3::Lrnr_glm_fast$new()
params_MONITOR <- sl3::Lrnr_glm_fast$new()
OData <- fitPropensity(OData,
            gform_CENS = gform_CENS, stratify_CENS = stratify_CENS, params_CENS = params_CENS,
            gform_TRT = gform_TRT, params_TRT = params_TRT,
            gform_MONITOR = gform_MONITOR, params_MONITOR = params_MONITOR)

## End(Not run)

# ----------------------------------------------------------------------
# Running TMLE based on the previous fit of the propensity scores.
# Also applying Random Forest to estimate the sequential outcome model
# ----------------------------------------------------------------------
## Not run: 
t.surv <- c(0:5)
Qforms <- rep.int("Qkplus1 ~ CVD + highA1c + N + lastNat1 + TI + TI.tminus1", (max(t.surv)+1))
models <- sl3::Lrnr_h2o_grid$new(algorithm = "randomForest",
                           ntrees = 100, learn_rate = 0.05, sample_rate = 0.8,
                           col_sample_rate = 0.8, balance_classes = TRUE)
tmle_est <- fit_TMLE(OData, tvals = t.surv, intervened_TRT = "TI.set1",
            Qforms = Qforms, models = models,
            stratifyQ_by_rule = TRUE)

## End(Not run)

## Not run: 
t.surv <- c(0:5)
Qforms <- rep.int("Qkplus1 ~ CVD + highA1c + N + lastNat1 + TI + TI.tminus1", (max(t.surv)+1))
models <- sl3::Lrnr_h2o_grid$new(algorithm = "randomForest",
                           ntrees = 100, learn_rate = 0.05, sample_rate = 0.8,
                           col_sample_rate = 0.8, balance_classes = TRUE)
tmle_est <- fit_TMLE(OData, tvals = t.surv, intervened_TRT = "TI.set1",
            Qforms = Qforms, models = models,
            stratifyQ_by_rule = FALSE)

## End(Not run)

osofr/stremr documentation built on Jan. 25, 2022, 8:07 a.m.