fit_cyclomort {cyclomort}

Estimate periodic hazard function.

Description

This function takes time-to-event data formatted as a cycloSurv object and estimates an underlying hazard function for a given number of seasons.

Usage

fit_cyclomort(
  x,
  inits = NULL,
  n.seasons = 2,
  method = "L-BFGS-B",
  period = NULL
)

Arguments

x

a cycloSurv object recording start and end times as well as status (dead/censored) and the length of one full period

inits

set of initial guesses; a named vector or list with values for "peak" and "duration". Leaving some or all of these parameters as NULL will trigger the automatic selection of an initial guess.

n.seasons

number of seasons to fit model to

method

method for optim call

period

expected periodicity of survival data. Can be passed in with cycloSurv input parameter

Value

a cmfit object containing parameter estimates for peaks, durations, and weights for each season

Examples

# Simulate data
T.morts1 <- simulate_cycloSurv(1000, period = 365, 
                             meanhazard = 0.3 / 365, 
                             peaks = c(0.25 * 365, 0.75 * 365), 
                             durations = c(0.3 * 365, 0.1 * 365), 
                             weights = c(0.7,0.3), 
                             plotme = FALSE)

# Estimate simulated data
fits <- fit_cyclomort(T.morts1, n.seasons = 2)
fits

# Plot results
plot(fits, nreps = 1000, monthlabs = TRUE)
# NB: `nreps` is for the bootstrap of the confidence interval 
# The default (5000) is slower but smoother

# Actual parameter values from simulated data
attributes(T.morts1)