Hybrid Format

Discrete-Time Multistate Modelling: New Analytical and Computational Tools (Part II)

Daniel Schneider
Laboratory of Population Health, February 14, 2023

Daniel Schneider from the Laboratory of Population Health gives a talk that builds on his previous one in June 2022 where the basic methodology of discrete-time multistate models was discussed. 

Abstract

This lab talk builds on his previous one (June 2022) where the basic methodology of discrete-time multistate models was discussed. After a quick recap, he will provide a progress update on methodological advancements and software implementation. As before, these improvements are related to the outcomes of state and overall durations (expectancies), mean age at first transition, and lifetime risk. A methodological advancement is the N-group comparison of outcomes, which will be illustrated using an empirical example related to cumulative (dis)advantage as described in Hale et al. (2022). The asymptotic calculation of covariance matrices in conjunction with the N-group comparison provide a statistical test for the presence of cumulative advantage or disadvantage.

A second focus of the talk will be a brief introduction to the Stata package -dtms- which implements a large number of discrete-time multistate procedures in a general fashion. A recent addition to that toolkit was the estimation of models with restricted state transitions, which includes disease progression models. Such restrictions on transitions are also important for applying the "trick" of expanding the state-space in order to relax the Markov assumption.

Register to Take Part

You would like to attend the Online Seminar Talk? You are very welcome. Please register by writing an e-mail to office-myrskyla@demogr.mpg.de.

Online Seminar Talk, February, 14th from 1 p.m. to 2 p.m. (Rostock time)