PhD in Advanced statistical methods for prognostic and treatment effect modeling

Which biomarkers can predict disease progression in Alzheimer’s disease? Or which factors predict treatment response in acute stroke patients or cancer? Are you going to help us solve these questions?

Recent advances in the analysis of big data allow efficient learning from datasets with many variables on a limited number of individuals. In parallel, there have been important advances in the field of causal inference. Building on these developments, this project aims to tailor regularization methods and causal inference techniques for medium dimensional prediction and treatment effect size estimation problems typical in medical settings, where limited amounts of data are typically collected at high cost. The project entails both methodological development and implementation, and aims to increase accessibility of state of the art methods to a wide audience.

Objectives include methodological developments and implementations that facilitate more effective prediction model development and evaluation in medium dimensional settings (e.g. where there is a clear tension between sample size and the number of model parameters). Challenges include both prognostic and treatment effect modeling. You will have access to high quality clinical data to inspire and implements methods. You will finalize the research with a PhD thesis.

The Big Statistics section on the Department of Epidemiology and Data Science has a strong tradition of applying novel mathematical insights into statistics applications for medical purposes. The proposed line of research into medium dimensional prediction also has a strong applied character. Key research lines in our section include: statistical omics, studying statistical methods for big p problems (e.g. gene, protein, metabolite, and exposome data); co-data learning: studying methods that allow incorporating external data or expert knowledge into statistical models; and causal inference. You will be co-supervised by dr. Jeroen Hoogland and prof. Mark van de Wiel from the department of Epidemiology and Data Science.