With an overarching goal of improving the way prognostic/ prediction models are used in routinely collected health data, Glen's research can be summarised into four pillars of research interests:
1) Understanding the mechanisms that drive and underpin observational data: this covers exploring why observational data are present (or missing) since the presence/absence of information might be informative of an individual’s health.
2) Optimising the ways in which risk models are developed. This focusses on statistical methods development to improve risk prediction modelling. Particular interests include multivariate (multi-outcome) risk prediction, penalisation methods and sample size.
3) Appropriate re-use of prediction models: how can we make better use of existing knowledge/ research in prognostic modelling? This includes both meta-analysis and validation studies
4) Apply existing and novel statistical methodologies into real-world clinical/applied investigations