ABSTRACT
The proportional hazards (PH) model is by far the most popular survival model in medical applications since its introduction in 1972 by Cox. A major selling argument of this model is the partial likelihood estimation technique which allows estimating regression parameters hereby ignoring the baseline hazard function. Numerous approaches have been suggested to fit a PH model to interval-censored data, from purely parametric methods to semiparametric methods. The chapter reviews the various flexible approaches and deals with the semiparametric approaches. It deals with the piecewise exponential baseline survival model (PEBS). This model assumes on the hazard scale a piecewise constant behavior with data-driven or user-defined knots. The PEBS model could be considered a first step towards nonparametric modelling of the baseline hazard function. The function frailtyPenal allows for fitting a PEBS PH model to interval-censored data.
