
Exposure–risk modeling for time-to-event data 135
When the exposure is driven by a complex popPK model, it is more con-
venient to c alculate H(t) numerically. NONME M has the fac ility to model
the popPK and TTE model jointly. The following is a s ample program for
a simple popPK model (adapted from Ho lford’s course notes for advanced
pharmacometrics at
http://holford.fmhs.auckland.ac.nz/teaching/pharmacometrics/advanced),
but it can be extended to any other popPK mo de ls specified in terms of O DE
(linear or nonlinear). The idea is to calculate c
∗
i
(t) and H
i
(t) by solving the
following ODEs with initial conditions H
i
(0) = 0 and c
∗
i
(0) = 0,
dc
∗
i
(t