Flexible parametric survival analysis
Royston-Parmar flexible parametric models as an alternative to Cox regression — flexible baseline hazards with straightforward, interpretable predictions.
What this course covers
Register interest
This course introduces flexible parametric (Royston-Parmar) models as an alternative to Cox regression for survival analysis. Rather than assuming a strict functional form for the baseline hazard, these models use restricted cubic splines to provide flexible estimation while enabling straightforward predictions and risk calculations.
The primary software is Stata (version 17+), using the stmerlin package, with full R solutions provided. A certificate is provided on 80%+ attendance.
Full course title: Flexible parametric survival analysis: An introduction to concepts, methods & software
Try it in your browser: the survival DGM explorer (Royston–Parmar & log-hazard splines) and the KM reconstruction & extrapolation tool (flexible-parametric fits to a lifetime horizon).
At a glance
What you'll learn
Splines & FPMs
Time-to-event data fundamentals, an introduction to splines, and flexible parametric survival models.
Royston-Parmar & stmerlin
The Royston-Parmar model and the spline-based Cox model, fitted with the stmerlin package in Stata.
Time-dependent effects
Modelling time-dependent effects and non-proportional hazards.
Predictions
Predictions from flexible parametric models and interpretable risk measures.
Conditional & marginal
Conditional and marginal predictions from a fitted model.
R solutions
Full R solutions to all practical exercises are provided alongside Stata.
Audience & prerequisites
Who it's for
Epidemiologists, statisticians, physicians, public health specialists or anyone with an interest in modelling time-to-event data.
Prerequisites
Basic epidemiology and biostatistics knowledge, and comfort with survival analysis models (Poisson and Cox regression). Basic Stata familiarity is assumed.