Bayesian evidence synthesis in practice



Christian Röver, University Medical Center Göttingen

Sebastian Weber, Novartis Pharma AG

Day: Sunday, September 6

Length of tutorial: full day

Technical level: medium


Meta-analytic methods are introduced starting from the general ideas (like exact and approximate likelihood approaches and hierarchical models) to more advanced issues (like prediction, shrinkage estimation and robustification). Motivated by medical applications, we show how information from historical studies may be used to derive a meta-analytic predictive (MAP) prior, which may then be utilized in a range of ways. In particular, we demonstrate applications making use of historical control data in clinical trials for the purpose of reducing the control group sample size, understanding their design properties, and evaluating the probability of success for a planned future trial. The tutorial includes hands-on exercises using R.

Local Organizer:

Institute of Biometry and Clinical Epidemiology
Charité - Universitätsmedizin Berlin & Berlin Institute of Health