In ddsjoberg/sjosmooth: Kernel smoothing for time to event data

knitr::opts_chunk$set(echo = TRUE)

Simulate Data

First, we'll simulated data with a changing beta coefficient.

library(dplyr)
library(ggplot2)
library(survival)
library(sjosmooth)

set.seed(23423)
n = 10000
dta_sim =
  dplyr::data_frame(
    x = rnorm(n),
    beta = case_when(
      x <= 0 ~ 0,
      x > 0 ~ 1
    ),
    xb = beta * x,
    time = -log(runif(n, 0, 1)) * exp(-xb),
    # getting the true survival probabilites at time 1 using x
    survt1 = exp(-exp(xb))
  )

ggplot(dta_sim, aes(x = x, y = survt1)) + 
  geom_line() +
  scale_y_continuous(limits = c(0, 0.5)) +
  labs(
    title = "True 1 Year Survival"
  )

sm_regression

Dashed line is the true beta coefficient, and the solid line is the estimated slope coefficient from the sm_regression() function.

ex1 = 
  sm_regression(
    dta_sim, 
    method = "coxph",
    formula = Surv(time) ~ x,
    weighting_var = "x",
    newdata = data_frame(x = seq(-3, 3, by = 0.2), time = 1)
  )

ex1 %>%
  mutate(
    tidy = purrr::map(model_obj, ~broom::tidy(.x, conf.int = TRUE))
  ) %>%
  filter(purrr::map_lgl(model_obj, ~!is.null(.x))) %>%
  select(newdata, tidy) %>%
  tidyr::unnest(newdata, tidy) %>%
  ggplot(aes(x = x, y = estimate)) +
  geom_line() +
  geom_ribbon(aes(ymin = conf.low, ymax = conf.high), alpha = 0.3) +
  geom_line(data = dta_sim %>% filter(x > -3 & x < 3), aes(x = x, y = beta), linetype = "dashed") +
  scale_y_continuous(limits = c(-0.2, 1.2))

sm_predict

Dashed line is the true 1 year survival, and the solid line is the estimated 1 year survival from the sm_predict() function.

ex2 = 
  sm_predict(
    dta_sim, 
    method = "coxph",
    formula = Surv(time) ~ x,
    newdata = data_frame(x = seq(-3, 3, by = 0.2), time = 1),
    type = "survival"
  )

ex2 %>%
  ggplot(aes(x = x, y = .fitted)) +
  geom_line() +
  geom_ribbon(aes(ymin = .fitted.ll, ymax = .fitted.ul), alpha = 0.3) +
  geom_line(data = dta_sim %>% filter(x > -3 & x < 3), aes(x = x, y = survt1), linetype = "dashed") +
  scale_y_continuous(limits = c(0, 0.5))