R/ModelRTS.R

In citronmeliss/predictionpackr: Predictionpackr

#' Model RTS
#'
#' This function makes predictions with the RTS model.
#' @param data The study data frame. No default.
#' @export
#'

ModelRTS <- function(data)
{
  ## Define model_variables
  model_variables <- c("sbp",
                       "gcs",
                       "rr")

  ## Define cut points
  cut_points <- list(sbp = c(0, 1, 49, 75, 89, Inf),
                     gcs = c(3,4,5,8,12,Inf),
                     rr = c(0,1,5,9,29, Inf))

  ## Define scores of variabels
  scores <- list(sbp = c("0", "1", "2", "3", "4"),
                 gcs = c("0", "1", "2", "3", "4"),
                 rr = c("0", "1", "2", "4", "3"))

  ## Define RTS coefficients
  RTS_coefficients <- c(0.9368,
                        0.7326,
                        0.2908)

  ## Use bin.model.variables to group model variables into scores
  binned_variables <- lapply(setNames(model_variables, nm = model_variables),
                             function(col){
                               BinModelVariables(data,
                                                 model_variables,
                                                 cut_points,
                                                 scores)[ ,col]
                             }
  )
  ## Apply RTS coefficients to variables and sum rows to generate predictions.
  ## Then, invert.
  predictions <- rowSums(mapply('*',
                                binned_variables,
                                RTS_coefficients))

  return (predictions)
}