(Log) time acceleration factors in semi-Markov models

Extracts the covariate effects on scale parameters in phase-type approximation sojourn distributions in semi-Markov models. Note these are not hazard ratios for transitions on the observable state space. They are time acceleration factors, such that an increase in one covariate unit makes time pass at taf times the speed, such that the sojourn time is expected to reduce by \(exp(-taf)\). The log TAFs are labelled \(\beta\) in the paper and vignettes.

Usage

logtaf(draws)

taf(draws)

Arguments

draws

Object returned by msmbayes.

Value

A data frame containing samples from the posterior distribution. See qdf for notes on this format and how to summarise.

taf returns these on the natural scale, logtaf returns them on the log scale.

Details

See loghr for effects of covariates on transitions out of Markov states, which are log hazard ratios.