R/estimate_mu0_t.R

In SLCARE: Semiparametric Latent Class Analysis of Recurrent Events

#' @title Estimate mu0(t)
#' @description Estimate mu0 with the Nelson-Aalen type estimator under the assumed multiplicative intensity modeling of recurrent events.
#' @param time_long long-format observed event time.
#' @param censor_long long-format censoring time (longest follow up time).
#' @param t time point of interest.
#' @return estimated mu0(t).
#' @noRd
mu_t <- function(time_long, censor_long, t) {
  # exp(t=0) = 1
  mu_0 <- 1
  # filter obs_time > t, remove 0 event patient since they do not contribute to this formula
  list <- !is.na(time_long)
  time_long <- time_long[list]
  censor_long <- censor_long[list]
  time_t <- time_long[time_long >= t]
  n_time_t <- length(time_t)

  if (n_time_t >= 1) {
    nn <- numeric(n_time_t)
    for (i in 1:n_time_t) {
      nn[i] <- sum((time_t[i] <= censor_long) * (time_t[i] >= time_long))
    }

    mu_0 <- exp(-sum(1 / nn))
  }

  if (mu_0 == 0) {
    mu_0 <- 1e-6
  }
  return(mu_0)
}