By Dr Ruth Keogh, London School of Hygiene and Tropical Medicine

Abstract: In ‘dynamic’ prediction of survival we make updated predictions of individuals’ survival over time as new information becomes available about their health status. Landmarking is an attractive and flexible method for dynamic prediction.

Large observational patient databases provide longitudinal data on clinical measurements and present opportunities to develop ‘personalised’ dynamic predictions of survival. This work is motivated by the aim of developing dynamic prediction models for survival in people with cystic fibrosis, one of the most common inherited life-shortening diseases, using the US Cystic Fibrosis Patient Registry, which contains longitudinal clinical data on over 40,000 people. I will discuss some of the challenges faced in making dynamic predictions of survival in this data and show how they can be addressed in the landmarking framework. I will also show some comparisons between landmarking and the alternative approach of joint modelling of longitudinal and survival data and hopefully convince you that landmarking has a number of advantages.