Get information and plots for flexsurvspline models — quick_fit

This function checks whether flexsurvspline models converge, by fitting the models via fit.models in the form survival::Surv(time,event) ~ 1 for each strata, and for those models without warnings/errors, it returns fit information and plots.

Usage

quick_fit_splines(
  data,
  time,
  event,
  strata,
  dists,
  times = NULL,
  strata_labels = NULL,
  formula = NULL,
  weights = NULL,
  xlab = "Time",
  font.family = "Roboto Condensed",
  plot.theme = theme_easysurv(),
  add_interactive_plots = FALSE,
  plot_predictions = "flexsurv",
  include_ci = TRUE,
  get_mean = FALSE
)

Arguments

data: A tibble or data frame containing the survival data with columns for time, event and strata.
time: The name of the time variable in data
event: The name of the event variable in data
strata: The name of the strata variable in data
dists: A data frame with columns knots and scale specifying the number of knots to be used and the type of spline model to be executed respectively. The possible models are: Proportional hazards model (scale = "hazard"), probit model (scale = "normal"), proportional odds model (scale = "odds").
times: Optional. A set of times at which to calculate predicted survival. The default is seq(from = 0, to = ceiling(max(data$time)*2.5), length.out = 200)
strata_labels: Optional. A character vector containing the names of the stratas (default is NULL). Provide in a consistent order with levels(as.factor(data$strata)).
formula: Optional. Surv() formula. Default is survival::Surv(time,event) ~ 1
weights: Optional for case weights. The function expects a string corresponding to a variable name within the data.
xlab: The x-axis label for plots. Default is "Time".
font.family: The name of the font for the plots. Default is "Roboto Condensed".
plot.theme: ggplot2 theme for the plots. Default is theme_easysurv().
add_interactive_plots: Optional. Whether to include plotly outputs for hazard and fit plots. Default is FALSE.
plot_predictions: Optional. Whether to plot the predictions using the flexsurv package or survHE package. Default is "flexsurv".
include_ci: Logical indicating whether to include confidence intervals in the survival probability predictions. Default is TRUE.
get_mean: Optional. Whether to attempt to calculate mean survival times for the fit_averages object. Defaults to FALSE.

Value

An object of class quick_fit_spline containing the following elements:

converged: A list of the distributions that converged
fits: The survHE::fit.models output
hazard_plots: Smoothed hazard plots for each strata
hazard_plotly: Interactive hazard plots (if add_interactive_plots = TRUE)
fit_plots: Plot of extrapolations over the specified times parameter
fit_plotly: Interactive fit plots (if add_interactive_plots = TRUE)
goodness_of_fit: AIC and BIC output, alongside ranks, for each distribution
surv_params: The flexsurv parameters for each model and their vcov matrix
fit_averages: The median, mean and restricted mean survival times for each distribution
predicted_fits: Predicted survival proportions over the times parameter, if provided

Examples

if (FALSE) {
# Load required packages
library(survival)

input_data <- survival::lung

surv_data <- tibble(
  time = input_data$time,
  event = input_data$status - 1,
  strata = as.factor(input_data$sex)
)

# Define distributions
spline_dists <- tibble(
  "knots" = c(1, 2, 3),
  "scale" = c("odds", "hazard", "normal")
)

fit_check_splines <- quick_fit_splines(
  data = surv_data,
  time = "time",
  event = "event",
  strata = "strata",
  dists = spline_dists
)
}