weightPredict: Continuous time weight estimation based on 'predict.aalen'.

In palryalen/ahw: Re-weighting point processes based on additive hazard models

Description

Extracts cumulative hazard estimates for each individual. Each individual receives a weight process evaluated at pre-specified time points. The weight process is estimated as a cumulative product involving estimated cumulative hazard increments, and a hazard ratio estimated using a smoothing parameter b.

Usage

weightPredict(fPred, cfPred, wtFrame, ids, eventTimes, eventIds, b)

Arguments

fPred, cfPred, wtFrame, ids, eventTimes, eventIds, b)

fPred	predict.aalen object of the factual fit
cfPred	predict.aalen object of the counterfactual fit
wtFrame	data.frame or data.table for the at risk individuals
ids	All individuals in the data set
eventTimes	Observed event times
eventIds	Individuals that experience the event
b	Smoothing parameter

Value

data.table

Author(s)

Pål Christie Ryalen <[e-mail]>

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 (dataFr, atRiskState, eventTimes) 
{
    sortedEventTimes <- sort(eventTimes)
    baseTable <- as.data.table(dataFr)
    baseTable$rowNumber <- 1:nrow(baseTable)
    baseTable[, `:=`(isAtRisk, 1 * (from.state %in% atRiskState))]
    baseTable[, `:=`(numRep, 0)]
    baseTable[isAtRisk == 1, `:=`(numRep, as.numeric(sum(from <= 
        eventTimes & to > eventTimes))), by = rowNumber]
    baseTable[, `:=`(numRep, numRep + 1)]
    baseTable[, `:=`(to.state, as.character(to.state))]
    Table <- baseTable[rep(1:nrow(baseTable), times = baseTable[, 
        numRep]), ]
    Table[, `:=`(putEventTimes, 0)]
    Table[numRep > 1, `:=`(putEventTimes, c(sortedEventTimes[sortedEventTimes >= 
        from[1] & sortedEventTimes < to[1]], to[1])), by = id]
    Table[numRep == 1, `:=`(putEventTimes, to)]
    Table[, `:=`(to, putEventTimes)]
    Table[, `:=`(from, c(from[1], to[-length(to)])), by = id]
    Table[, `:=`(to.state, c(rep(0, length(to) - 1), baseTable$to.state[rowNumber[1]])), 
        by = rowNumber]
    return(Table)
  }

palryalen/ahw documentation built on Sept. 29, 2021, 6:20 p.m.