SurvSplit {eha}

Split a survival object at specified durations.

Description

Given a survival object, (a matrix with two or three columns) and a set of specified cut times, split each record into multiple subrecords at each cut time. The new survival object will be in ‘counting process’ format, with an enter time, exit time, and event status for each record.

Usage

SurvSplit(Y, cuts)

Arguments

Y

A survival object, a matrix with two or three columns.

cuts

The cut points, must be strictly positive and distinct.

Value

A list with components

Y

The new survival object with three columns, i.e., in 'counting process' form.

ivl

Interval No., starting from leftmost, (0, cuts[1]) or similar.

idx

Row number for original Y row.

Note

This function is used in phreg for the piecewise constant hazards model. It uses age.window for each interval.

Author(s)

Göran Broström

See Also

survSplit, age.window.

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function(Y, cuts){
    if (NCOL(Y) == 2) Y <- cbind(rep(0, NROW(Y)), Y)
    indat <- cbind(Y, 1:NROW(Y), rep(-1, NROW(Y)))
    colnames(indat) <- c("enter", "exit", "event", "idx", "ivl")
    n <- length(cuts)
    cuts <- sort(cuts)
    if ((cuts[1] <= 0) || (cuts[n] == Inf))
        stop("'cuts' must be positive and finite.")
    cuts <- c(0, cuts, Inf)
    n <- n + 1
    out <- list()
    indat <- as.data.frame(indat)
    for (i in 1:n){
        out[[i]] <- age.window(indat, cuts[i:(i+1)])
        out[[i]]$ivl <- i
        out[[i]] <- t(out[[i]])
    }
    Y <- matrix(unlist(out), ncol = 5, byrow = TRUE)
    colnames(Y) <- colnames(indat)
    list(Y = Y[, 1:3],
         ivl = Y[, 5],
         idx = Y[, 4]
         )
  }