About me
I am a Reader (equivalent to an Associate Professor outside the UK) in the School of Mathematics at the University of Edinburgh, where I am currently Director of Teaching.
My time is split between research, teaching, and administration — and I actually enjoy all three. My research may not — or rather, will not — change the world, but I firmly believe that we can change lives through education that is free or, at the very least, accessible.
Research
I work in biostatistics, mainly on the statistical evaluation of biomarkers and diagnostic tests: how well they distinguish between disease stages, and how factors such as age and gender change that accuracy. I use both Bayesian and frequentist (nonparametric) methods, with a preference for the former.
I developed some of the first methods for inference about several summary measures of diagnostic accuracy — the partial area under the ROC curve, the Youden index and the covariate-adjusted ROC curve — that make no assumptions about the distribution of test outcomes in each population, or about how covariates relate to those outcomes. I also introduced the first measure of diagnostic accuracy that accounts for a functional (curve-valued) covariate (Biostatistics, 2012).
More recently, with colleagues, I proposed the underlap coefficient, a measure of how well a biomarker distinguishes between more than two disease stages (Biometrics, 2026).
Many of these methods are freely available in the R package ROCnReg. More details are on the Publications and Software pages.