Dynamic survival prediction combining landmarking with a machine learning ensemble: Methodology and empirical comparison with standard techniques
Three talks on modern methods for clinical prediction modelling. This is the third in the series, and will be focusing dynamic prediction models that provide predicted survival probabilities and can be updated as new biomarker levels and test results become available can assist in this endeavour.
Two techniques for dynamic survival prediction with longitudinal data dominate the statistical literature: joint modelling and landmarking. We will present an approach that combines landmarking with a machine learning ensemble - the Super Learner. The ensemble combines predictions from different machine learning and statistical algorithms with the goal of achieving improved performance.
The proposed approach exploits discrete time survival analysis techniques to enable use of machine learning algorithms for binary outcomes. The methods are illustrated and compared using longitudinal data from the UK Cystic Fibrosis Registry.
Kamaryn Tanner is a PhD candidate at LSHTM in the Department of Medical Statistics and is supervised by Ruth Keogh, Linda Sharples and Rhian Daniel.
Please note that the time listed is Greenwich Mean Time (GMT)