R/run_projections.R

In kravitz-eli/simCondSurv: What the Package Does (One Line, Title Case)

#' Title
#'
#' @param simulated_times
#' @param crit_values
#'
#' @return
#' @export
#'
#' @examples
get_projections <- function(simulated_times, crit_values) {
  time_to_events <- sapply(simulated_times, function(x) x$censor_time[[1]])

  HR_at_events <- vector("numeric", length(simulated_times))
  post_pred_median_ctrl <- vector("numeric", length(simulated_times))
  post_pred_median_trt <- vector("numeric", length(simulated_times))

  for (i in seq_along(simulated_times)) {
    coxPH_model <- survival::coxph(survival::Surv(time, event) ~ trt, data = simulated_times[[i]])
    HR_at_events[[i]] <- exp(unname(coxPH_model$coefficients))

    KM_fit <- survival::survfit(survival::Surv(time, event) ~ trt, data = simulated_times[[i]])

    # Get the median survival times for the control and experiment arm
    post_pred_median_ctrl[[i]] <- survival:::quantile.survfit(KM_fit, probs = 0.50)$quantile[1]
    post_pred_median_trt[[i]] <- survival:::quantile.survfit(KM_fit, probs = 0.50)$quantile[2]
  }


  prob_reject <- sapply(crit_values, function(x) mean(HR_at_events < x))


  return(list(
    "time_to_events" = time_to_events,
    "HR_at_events" = HR_at_events,
    "post_pred_median_ctrl" = post_pred_median_ctrl,
    "post_pred_median_trt" = post_pred_median_trt
  ))
}