Compute day- and covariate- specific cumulative incidences
Source:R/compute_psi_dx.R
compute_psi_dx_t0.RdWrapper that calls internal functions for predicting cumulative incidences from
fitted hazard models. Returns the predicted exposure-specific cumulative
incidences side by side for each time- and covariate- pair in newdata.
Arguments
- fit_0
A fitted model returned from
fit_model_0()- fit_1
A fitted model returned from
fit_model_1()- exposure_time
Name of the time-to-exposure variable in
newdata. Used to compute \(\psi_0(t_0; d,x)\) where \(d + \tau\) and \(d + t_0\) are needed.- t0
Time since exposure at which to evaluate cumulative incidence.
- tau
Delay period
- newdata
New data at which to do predictions.
Value
A data frame with one row per row of newdata with the predicted
cumulative incidences and component survival probabilities:
psi_0_dx,surv_0_d_plus_tau,surv_0_d_plus_t0psi_1_dx,surv_1_tau,surv_1_t0
Details
Definitions of the cumulative incidences returned: $$\psi_0(t_0; d,x) = 1 - S_0(d+t_0; x)\,/\,S_0(d+\tau; x)$$ $$\psi_1(t_0; d,x) = 1 - S_1(t_0; d,x)\,/\,S_1(\tau; d,x)$$
where \(d\) represents exposure time, \(x\) represents covariates, and \(S_v\) represents the survival probability from hazard model for exposure \(v\).