Short Course

PD Dr. Fabian Scheipl and Dr. Andreas Bender will provide a short course on a specific model class useful for flexible time-to-event modeling, namely the so-called Piece-wise exponential (additive) models (PEMs / PAMs).

The short course will take place on Sunday, Juli 16, 2023, and will take about 6 hours. Registration for the short course will be open until May 31, 2023, and the fee is 120€.

Course description

Piece-wise exponential (additive) (mixed) models (PEM/PAMM) are highly flexible and performant models for time-to-event data, including models with time-varying effects, time-dependent covariates, frailties and models for left-truncated data, recurrent events and competing risks. PAMMs can be represented as generalized additive (mixed) models and can therefore be estimated using GAM software (e.g. mgcv) or, even more generally, any supervised learning algorithm able to optimize a  Poisson likelihood. This offers huge benefits in terms of flexibility w.r.t. to the specification of covariate effects (e.g. non-linear, time-varying, cumulative and/or random effects) as well as the choice of inferential framework (Bayesian, (penalized) likelihood, boosting, etc).

This short course will introduce the required theoretical background (it's simple!), followed by lots of demonstrations and hands-on exercises with supplied data and R code to work through some practical applications using the presenters' own {pammtools} package.

Speaker profiles

After a PhD in Bayesian Statistics and postdoctoral work on functional data analysis, Fabian Scheipl is now a lecturer at the Department of Statistics at LMU Munich and a PI at the Munich Center for Machine Learning. His most recent research is focused on improving unsupervised methodology for functional data.  He has worked on PAMMs, their implementation and their application to challenging clinical data sets for many years.

Andreas Bender is a postdoctoral researcher and lecturer at the Department of Statistics at LMU Munich and senior consultant at the Statistical Consulting Unit. After obtaining a PhD in Statistics, he was a Postdoc at the Big Data Institute, University of Oxford, and Interim Professor at the Institute of Statistics, University of Ulm. Currently, he is Coordinator of the Machine Learning Consulting Unit at the Munich Center for Machine Learning and is junior group leader on machine learning survival analysis. He started to work on PAMMs during his PhD, and worked on this model class as well as survival analysis in general ever since, including extensions to machine and deep learning.