R/Param_TSM.R

In lurui0421/TMLE-for-causal-inference-: The Extensible TMLE Framework

#' Treatment Specific Mean
#'
#' Parameter definition for the Treatment Specific Mean (TSM): $E_W[E_{Y|A}(Y|A=a|W)|$. Currently supports multiple static intervention nodes.
#' Does yet not support dynamic rule or stochastic interventions.
#'
#' @section Current Issues:
#' \itemize{
#'   \item clever covariates doesn't support updates; always uses initial (necessary for iterative TMLE, e.g. stochastic intervention)
#'   \item doesn't integrate over possible counterfactuals (necessary for stochastic intervention)
#'   \item clever covariate gets recalculated all the time (inefficient)
#' }
#' @importFrom R6 R6Class
#' @importFrom uuid UUIDgenerate
#' @importFrom methods is
#' @family Parameters
#' @keywords data
#'
#' @return \code{Param_base} object
#'
#' @format \code{\link{R6Class}} object.
#'
#' @section Constructor:
#'   \code{define_param(Param_TSM, observed_likelihood, intervention_list, ..., outcome_node)}
#'
#'   \describe{
#'     \item{\code{observed_likelihood}}{A \code{\link{Likelihood}} corresponding to the observed likelihood
#'     }
#'     \item{\code{intervention_list}}{A list of objects inheriting from \code{\link{LF_base}}, representing the intervention.
#'     }
#'     \item{\code{...}}{Not currently used.
#'     }
#'     \item{\code{outcome_node}}{character, the name of the node that should be treated as the outcome
#'     }
#'     }
#'

#' @section Fields:
#' \describe{
#'     \item{\code{cf_likelihood}}{the counterfactual likelihood for this treatment
#'     }
#'     \item{\code{intervention_list}}{A list of objects inheriting from \code{\link{LF_base}}, representing the intervention
#'     }
#' }
#' @export
Param_TSM <- R6Class(
  classname = "Param_TSM",
  portable = TRUE,
  class = TRUE,
  inherit = Param_base,
  public = list(
    initialize = function(observed_likelihood, intervention_list, ..., outcome_node = "Y") {
      super$initialize(observed_likelihood, ..., outcome_node = outcome_node)
      private$.cf_likelihood <- make_CF_Likelihood(observed_likelihood, intervention_list)
    },
    clever_covariates = function(tmle_task = NULL, fold_number = "full") {
      if (is.null(tmle_task)) {
        tmle_task <- self$observed_likelihood$training_task
      }
      intervention_nodes <- names(self$intervention_list)
      pA <- self$observed_likelihood$get_likelihoods(tmle_task, intervention_nodes, fold_number)
      cf_pA <- self$cf_likelihood$get_likelihoods(tmle_task, intervention_nodes, fold_number)

      HA <- cf_pA / pA

      # collapse across multiple intervention nodes
      if (!is.null(ncol(HA)) && ncol(HA) > 1) {
        HA <- apply(HA, 1, prod)
      }
      return(list(Y = unlist(HA, use.names = FALSE)))
    },
    estimates = function(tmle_task = NULL, fold_number = "full") {
      if (is.null(tmle_task)) {
        tmle_task <- self$observed_likelihood$training_task
      }

      # todo: extend for stochastic
      cf_task <- self$cf_likelihood$cf_tasks[[1]]


      Y <- tmle_task$get_tmle_node(self$outcome_node)


      # clever_covariates happen here (for this param) only, but this is repeated computation
      HA <- self$clever_covariates(tmle_task, fold_number)[[self$outcome_node]]

      # clever_covariates happen here (for all fit params), and this is repeated computation
      EYA <- unlist(self$observed_likelihood$get_likelihood(tmle_task, self$outcome_node, fold_number), use.names = FALSE)

      # clever_covariates happen here (for all fit params), and this is repeated computation
      EY1 <- unlist(self$cf_likelihood$get_likelihood(cf_task, self$outcome_node, fold_number), use.names = FALSE)

      # todo: separate out psi
      # todo: make this a function of f(W)
      psi <- mean(EY1)
      IC <- HA * (Y - EYA) + EY1 - psi

      result <- list(psi = psi, IC = IC)
      return(result)
    }
  ),
  active = list(
    name = function() {
      param_form <- sprintf("E[%s_{%s}]", self$outcome_node, self$cf_likelihood$name)
      return(param_form)
    },
    cf_likelihood = function() {
      return(private$.cf_likelihood)
    },
    intervention_list = function() {
      return(self$cf_likelihood$intervention_list)
    },
    update_nodes = function() {
      return(self$outcome_node)
    }
  ),
  private = list(
    .type = "TSM",
    .cf_likelihood = NULL
  )
)