Statistical regression models are central to many fields of quantitative research. A regression analysis typically starts with model building, an activity with which medical researchers will have some familiarity. A particular set of modern statistical tools, known as objective Bayesian model selection, is able to support the model building process and to enhance the suitability of regression models for a wider range of research applications. Until now, these tools have only been available for one type of regression model, the linear regression model.
The current SNF-funded project will allow Professor Held and his team to expand objective Bayesian model selection methods to other types of regression models, namely the generalized linear model. This work will strengthen many of the statistical techniques currently used in biomedical research. These tools and methods will be key in the field of personalized medicine, helping researchers better predict the risk of disease outcomes.